lester

							             A DEFENSE OF PAID FAMILY LEAVE


                                Gillian Lester∗



                                  I. Introduction

      The problem of combining work and family life is perhaps the central
challenge for the contemporary American family. In this Article, I evalu-
ate and defend government provision of paid family leave, a beneªt that
would allow workers to take compensated time off from work for purposes
of family caregiving.
      A legal intervention in the arena of work-family accommodation can
only build on some prior normative understanding of the family, and em-
bedded within that, contested value choices about women’s identities and
entitlements in workplace, family, and society. I am not the ªrst legal scholar
to advocate paid family leave of some kind.1 The additional contribution

    ∗ Professor of Law, UCLA School of Law. I am grateful for the comments of Rick Abel,

Grace Blumberg, Doriane Coleman, Mark Kelman, Chris Littleton, Ruth Milkman, Vicki
Schultz, Michael Selmi, Seana Shiffrin, Kirk Stark, Lynn Stout, Eric Talley, Rip Verkerke,
Adam Winkler, Jonathan Zasloff, and Noah Zatz, workshop participants at Cornell, Duke,
Minnesota, Texas, UCLA, and Washington University law schools, the research assistance of
Xia Chen, Paul Foust, Kevin Gerson, Nikki Hollings-worth, Stephanie Hwang, Cheryl Kelly,
Michael Kovaleski, Arusi Loprinzi, Eileen O’Brien, and Dil Parkinson, and funding by the
Sloan Foundation, the UCLA Academic Senate, and the UCLA School of Law Dean’s Fund.
    1 See, e.g., Jeremy I. Bohrer, You, Me, and the Consequences of Family: How Federal Law

Prevents the Shattering of the “Glass Ceiling,” 50 Wash. U. J. Urb. & Contemp. L. 401,
418–21 (1996); Erin P. Drew, The Birth and Adoption Unemployment Compensation Ex-
periment: Did the Department of Labor Go Too Far?, 106 Dick. L. Rev. 367, 387–88
(2001); Arline Friscia, Reºections on Legislation: The Worker-Funded Leave Act: The Time
is Now To Help Build Stronger Families with a More Stable Economy, 26 Seton Hall
Legis. J. 73, 76–84 (2001); Mikel Glavinovich, International Suggestions for Improving
Parental Leave Legislation in the United States, 13 Ariz. J. Int’l & Comp. L. 147, 167–74
(1996); Arielle Horman Grill, The Myth of Unpaid Family Leave: Can the United States
Implement a Paid Leave Policy Based on the Swedish Model?, 17 Comp. Lab. L.J. 373,
383–90 (1996); Emily A. Hayes, Bridging the Gap Between Work and Family: Accomplish-
ing the Goals of the Family and Medical Leave Act of 1993, 42 Wm. & Mary L. Rev.
1507, 1532–38 (2001); Samuel Issacharoff & Elyse Rosenblum, Women and the Work-
place: Accommodating the Demands of Pregnancy, 94 Colum. L. Rev. 2154, 2214–21
(1994); Donna Lenhoff & Claudia Withers, Implementation of the Family and Medical
Leave Act: Toward the Family-Friendly Workplace, 3 Am. U. J. Gender & L. 39, 53–54
(1994); Michael Selmi, Family Leave and the Gender Wage Gap, 78 N.C. L. Rev. 707,
770–73 (2000); Stephen D. Sugarman, Short Term Paid Leave: A New Approach to Social
Insurance and Employee Beneªts, 75 Cal. L. Rev. 465, 467–76 (1987); Katherine Eliza-
beth Ulrich, Insuring Family Risks: Suggestions for a National Family Policy and Wage
Replacement, 14 Yale. J.L. & Feminism 1, 16–68 (2002); Angie K. Young, Assessing the
Family and Medical Leave Act in Terms of Gender Equality, Work/Family Balance, and the
Needs of Children, 5 Mich. J. Gender & L. 113, 154 (1998).
2                        Harvard Journal of Law & Gender                            [Vol. 28

here is to offer a normative defense of such a program based on its poten-
tial to increase the workforce participation of those who bear the princi-
pal obligation of caregiving—women. This, I argue, will increase equality
of economic opportunity and the distribution of social power associated
with status in paid labor markets. It also will enhance women’s capacity
to determine the conditions of their lives. In advocating paid family leave, I
distinguish myself from those who would make family care subsidies avail-
able equally to caregivers who do and do not participate in the paid work-
force, and from those who would shun workplace accommodations in favor
of more “commodiªed” caregiving institutions external to the family.
      Paid family leave is particularly valuable, I argue, because other possi-
ble alternatives, such as daycare, cannot entirely replicate the value of per-
sonal time away from work to engage directly in family caregiving. For
women currently working who want to give personal care to family mem-
bers but cannot afford adequate time off to do so, paid family leave will
improve their quality of life and beneªt those they care for. For women
on the margin between working and staying home, the availability of paid
leave may make market work more feasible and attractive, and as a result,
increase their attachment to the workforce. At the same time, we must be
wary of overly generous leave provision. Very generous leave provisions
might encourage such lengthy absences from the job as to undermine
women’s development of human capital and connection to the workforce.
Further, the method used to ªnance the program must be sensitive to im-
portant issues of distributive justice and the challenge of ensuring that
the program confers gains on its intended beneªciaries. The government
should spread at least some of the costs of the program beyond those work-
ers—women in their childbearing years—most likely to take leave.
      Paid family leave would have two components. It would have a fam-
ily illness leave component, i.e., temporary paid leave for someone who
is not herself incapacitated, but who has a familial obligation to another
person who is seriously ill or disabled. It would also have a parental
leave component, covering non-medical temporary leave for purposes of
allowing parents to nurture newborn children. The Family and Medical
Leave Act (“FMLA”) mandates that employers give up to twelve weeks
of job-protected leave per year to workers who need to care for a new-
born child or their own serious illness or the illness of a family member.2
Coverage limitations mean that only about half of all workers, and less than
one-third of steadily employed new mothers, receive these protections.3


    2 Family and Medical Leave Act of 1993, 29 U.S.C. §§ 2601–2654 (2000). Note that

the FMLA provides both (unpaid) family and personal illness leaves. The analysis in this
Article is limited to family leave. This is not because paid personal illness leave is unim-
portant, but because it raises a conceptually distinct set of questions that are not my central
focus.
    3 Christopher J. Ruhm, Policy Watch: The Family and Medical Leave Act, 11 J. Econ.

Persp. 175, 177 (1997). “Steadily employed,” as used here, means employed for at least
2005]                 A Defense of State Paid Family Leave                                   3

More importantly, the law does not require wage replacement. This makes
the American system the least generous of industrialized nations. All west-
ern European nations have programs that give women workers the right
to at least three months paid maternity leave, with as much as a year or more
in some countries, as well as paid parental leave—for either parent.4
      For workers who need to take time off to address family or medical
needs, ªnancial worries loom largest among their anxieties about taking
leave.5 The hardship of lost wages leads some workers to foreshorten their
time away from work6 or simply forgo a needed leave.7 A sizeable per-
centage of workers who lack access to paid beneªts resort to public assis-
tance for support during family leaves.8 Finally, although more difªcult
to measure, there are likely some workers who would enter or remain in the
workforce if there were better prospects of supported family leaves, but
who instead quit or stay home to address their family or medical needs.
      Recently, the debate over paid family leave has been revitalized. In
the past few years, twenty-one states have introduced bills to expand their
unemployment insurance (“UI”) programs to provide wage replacement
to parents following the birth or adoption of a child.9 In addition, several
states are considering bills that would expand existing temporary disabil-
ity insurance (“TDI”) programs or create new public insurance schemes



one year before childbirth. Id. The Act only covers employees who have worked for their
employer for twelve months and 1250 hours in the previous year, and whose employer has
ªfty or more employees working within seventy-ªve miles of the worksite. 29 U.S.C.
§§ 2611(2)(A)–(B).
     4 Ruhm, supra note 3, at 176.
     5 U.S. Dep’t of Labor, FMLA Survey: Balancing the Needs of Families and

Employers, at tbl.4.1 (2001) [hereinafter Balancing the Needs] (reporting that among
worries expressed by workers who took a leave in 2000 to care for a newborn child or their
own or a family member’s serious illness, whether or not covered by the FMLA, not hav-
ing enough money to cover their basic needs was cited most frequently (53.8% of leave-
takers)), available at http://www.dol.gov/asp/fmla/toc.htm (last visited Nov. 18, 2004).
     6 Id. at tbl.4.8 (reporting that 37% of leave-takers in 2000 reported cutting short their

leave time to cope with the hardship of lost wages).
     7 Although 16.5% of all employees in the United States took leaves of absence from

work to handle family or medical needs in 2000, another 2.4% of workers did not take
leave despite reporting that they needed it (i.e., roughly 13% of workers who needed to
take a leave did not take it). Id. at tbls.2.1, 2.14. Among those who were unable to take a
needed leave, the most common reason cited (77.6%) was not being able to afford it. Id. at
tbl.2.17. Workers who take leaves generally are more educated, have higher incomes, and
are more likely to earn a salary rather than an hourly wage than those who do not. Id. § 2.1.3.
     8 Id. at tbl.4.8 (reporting that 8.7% of leave-takers in 2000 used public assistance to

ªnance leaves).
     9 These states include Arizona, Florida, Hawaii, Illinois, Indiana, Kansas, Louisiana,

Maryland, Massachusetts, Minnesota, Mississippi, Missouri, Nebraska, New Jersey, New
Mexico, New York, Oregon, Pennsylvania, Texas, Vermont, and Washington. See Nat’l P’ship
for Women & Families, State Paid Leave Initiatives in the 2004 and Prior State
Legislatures: Making Family Leave More Affordable (2004) [hereinafter Leave Ini-
tiatives] (reporting updates of state legislative developments in paid family leave), avail-
able at http://www.nationalpartnership.org/portals/p3/library/paidleave/stateroundup2004.pdf
(last visited Nov. 18, 2004).
4                        Harvard Journal of Law & Gender                           [Vol. 28

to provide paid parental or family illness leave.10 In 2002, California be-
came the ªrst state in the nation to provide employees paid leave beneªts
not only for personal illness (including maternity leave), but also for pa-
rental bonding and caring for sick family members.11 These developments
make closer examination of paid family leave timely.
     Although this Article deals exclusively with paid family leave policy,
the goal of equalizing men’s and women’s respective contributions to both
market work and family caregiving can only be achieved through a com-
posite of interlocking social policies. For example, affordable, high-quality,
publicly available childcare, while not a substitute for paid leave, is a critical
part of the picture.12 Effective antidiscrimination laws,13 income tax poli-
cies that do not penalize dual-income married couples,14 and a shorter

     10 E.g., A.B. 173, 2004 Leg., 211th Sess. (N.J. 2004) (proposing “Family Leave Insur-

ance,” which would expand existing TDI and UI programs to cover family illness and
birth/adoption leaves, respectively, to be paid for by a payroll tax on workers); H.B. 2399,
2004 Leg., 58th Sess. (Wash. 2004) (would create “Family Leave Insurance,” which would
pay employees a ºat-rate weekly payment of $250 ªnanced by taxing employers and em-
ployees one cent each per work hour); S.B. 6272, 2004 Leg., 58th Sess. (Wash. 2004)
(same); H.B. 25, 2003 Leg., 22d Sess. (Haw. 2003) (would establish “Family Leave Beneªts
Insurance” to be ªnanced by a payroll contribution by employees and employers totaling a
maximum of two cents per work hour); S.B. 772, 2003 Leg., 22d Sess. (Haw. 2003)
(same); S.B. 778, 2003 Leg., 22d Sess. (Haw. 2003) (would expand Hawaii’s existing TDI
program to include parental and family illness leaves); H.P. 567, 2003 Leg., 121st Sess.
(Me. 2003) (proposing “Family Security Fund,” which provides between fourteen and twenty-
eight weeks paid maternity leave ªnanced by up to ninety cents employee deduction per
week plus matched employer contribution); S.P. 389, 2003 Leg., 121st Sess. (Me. 2003)
(would establish “Temporary Disability and Family Leave Beneªts Program,” which would
provide insurance covering leaves taken for birth, adoption, and family illness ªnanced by
equal contributions from employee and employer).
     11 California Family Temporary Disability Insurance Program, Cal. Unemp. Ins. Code

§ 3301 (Deering 2004) (amending state disability compensation program, which previously
provided compensation for individuals unable to work due to their own temporary illness
or disability, including pregnancy and childbirth, to also include up to six weeks of com-
pensation for leaves to care for an ill family member, or the birth, adoption, or foster care
placement of a new child).
     12 On public childcare reform, see, for example, Janet C. Gornick & Marcia K.

Meyers, Families That Work: Policies for Reconciling Parenthood and Employ-
ment 185–235 (2003); Deborah L. Rhode, Justice and Gender 129–31 (1989).
     13 On antidiscrimination litigation in the area of work-family conºict, see, for example,

Joan C. Williams & Nancy Segal, Beyond the Maternal Wall: Relief for Family Caregivers
Who Are Discriminated Against on the Job, 26 Harv. Women’s L.J. 77 (2003).
     14 See, e.g., Edward McCaffery, Taxing Women 16–23 (1999); Grace Blumberg,

Sexism in the Code: A Comparative Study of Income Taxation of Working Wives and Moth-
ers, 21 Buff. L. Rev. 49 (1971); Lawrence Zelenak, Marriage and the Income Tax, 67 S.
Cal. L. Rev. 339, 365–77 (1994) (showing how income tax laws are biased against secon-
dary earners—overwhelmingly women—in two-earner families, thereby creating disincen-
tives for women to choose market labor over unpaid home labor). Although the 2001 tax
reforms reduced the so-called “marriage penalty,” federal income tax treatment still creates
incentives for second earners to opt out of the labor market. See Jamie Heller, How New
Tax Law Relieves Marriage Penalty, Wall St. J., June 4, 2003, at D2 (explaining how
despite recent amendments to the federal tax code that purport to phase out the marriage
penalty, the poorest and wealthiest families get less than the full beneªt due to the limita-
tion of the amendments to the ªfteen percent tax bracket, and highlighting the persisting
marriage penalties in the earned income tax credit, phaseouts, capital loss offset provisions,
2005]                 A Defense of State Paid Family Leave                                 5

workweek15 are also key components. In addition, any paid leave policy
we adopt must contain effective incentives for men to take leaves, a chal-
lenge I pursue in some detail below.16 Thus, paid leave is not the only way
to advance the goal of greater gender equality in the balance of work and
family responsibilities, and paid leave cannot achieve such equality alone.
Nevertheless, it is a crucial piece of the puzzle, deserving extended reºec-
tion in its own right.
     The Article begins, in Part II, by analyzing why private markets do
not (for the most part) provide paid family leave insurance. In Part III, I
present my argument that the goal of increasing women’s workforce at-
tachment is a defensible basis on which to justify state intervention to create
paid family leave, and to ªnance it in a way that spreads its costs across
society. In Part IV, I turn to theoretical predictions and empirical ªndings
on the labor market effects of paid family leave programs. Part V pre-
sents objections to government subsidized paid family leave. Part VI ana-
lyzes different methods of beneªt ªnancing and explains how cost-
shifting affects who actually pays for a beneªt. In Part VII, I explore re-
form options. I discuss some general features a paid leave program ought
to have and illustrate with some examples of implementation models.

           II. State Intervention at the Work/Family Nexus

     The state might aim to reinforce various models of the contemporary
family, employing different policy instruments to achieve a variety of
goals.17 Thus, for example, the state might aim to increase national birth-
rates.18 It could do this by paying direct ºat-rate cash beneªts (not de-



and other areas of the tax code).
     15 On shortening or increasing the ºexibility of work hours, see Gornick & Meyers,

supra note 12, at 147–84; Joan Williams, Unbending Gender: Why Family and Work
Conºict and What To Do About It 100 (2000) (reviewing ethnographic studies de-
scribing working-class women’s anxiety about leaving children in low-quality daycare and
the strains of housework and the “split shift” on marital and family life); Vicki Schultz,
Life’s Work, 100 Colum. L. Rev. 1881, 1956–57 (2000) [hereinafter, Schultz, Life’s Work];
Vicki Schultz & Allison Hoffman, Precarious Work and Working Time: The Case for a
Reduced Workweek in the United States (to be published in yet untitled book, Judy Fudge
& Rosemary Owens eds., forthcoming 2005); Belinda M. Smith, Time Norms in the Work-
place: Their Exclusionary Effect and Potential for Change, 11 Colum. J. Gender & L.
271, 357–58 (2002).
     16 I discuss incentives for men to take leaves in more detail in Part VII, infra.
     17 A helpful overview of the issues discussed in this paragraph, in the context of Euro-

pean social policy experiments, can be found in Sheila B. Kamerman & Alfred J. Kahn,
Trends, Issues, and Possible Lessons, in Child Care, Parental Leave, and the Under
3s: Policy Innovation in Europe 201 (Sheila B. Kamerman & Alfred J. Kahn eds.,
1991) [hereinafter Child Care, Parental Leave, and the Under 3s].
     18 See id. at 204 (identifying France, Germany, and Hungary as having explicitly pro-

natalist policies); Anne Hélène Gauthier & Jan Hatzius, Family Beneªts and Fertility: An
Econometric Analysis, 51 Pop. Stud. 295, 295 (1997) (citing France and Luxembourg as
examples of countries that have in recent years adopted explicitly pronatalist policies).
6                       Harvard Journal of Law & Gender                          [Vol. 28

pendent on wage income) to families with children.19 Or, the state might
seek to recognize and reinforce a particular conception of “family values”
by supporting traditional single-earner husband-wife families.20 It could
do so by skewing beneªts toward married couples with children that have
one primary caregiver who is not working full time, or by providing sub-
sidies designed to create equivalency between paid work and work in the
home.21 The state might instead aim to encourage women’s workforce par-
ticipation, through such policies as public daycare.22 Yet another goal might
be to support good child development.23
     In short, government policy can inºuence the family form by shifting
the relative economic rewards attached to competing paths of behavior.
This is so even for policies neutrally described as endeavoring to “expand
choice”: if a state intervention makes a choice previously unavailable to
some portion of the population ªnancially more attractive, the effect will be
to encourage individuals on the margin toward the new option. As an ex-
ample, the work-family beneªts policies of the past ªfteen years in the
former Federal Republic of Germany, touted expressly by the govern-
ment as advancing the principle of free choice—i.e., that people rearing
children should have a choice between working and staying home—created
multiple ªnancial incentives for mothers to remain in or return to the home
rather than work.24 This puts before us the question of the social objec-
tives that might be achieved through paid family leave policies.

               A. Current Wage Replacement for Family Leaves

     Currently, there are limited sources of ªnancial support—aside from
savings or spousal earnings—for workers who take job leaves for child-
rearing or family caregiving purposes. American families with children


     19 See Gauthier & Hatzius, supra note 18, at at 295 (ªnding that in a study between

1970 and 1990 of family policies in twenty-two member countries of the Organisation for
Economic Co-Operation and Development, cash beneªts had a positive but limited effect
on fertility).
     20 Kamerman & Kahn, supra note 17, at 205 (citing Austria and Germany as having

this objective).
     21 Lilja Mósesdóttir, The Interplay Between Gender, Markets and the State

in Sweden, Germany and the United States 54 (2001) (describing German legislation
over the past ªfteen years giving incentives to mothers to interrupt their labor force par-
ticipation to care for babies and toddlers as well as elders and to work part time while
raising older children); Christiane Schiersmann, Germany: Recognizing the Value of Child
Rearing, in Child Care, Parental Leave, and the Under 3s, supra note 17, at 51, 63–
67 (giving examples of German policies established in the mid-1980s that reward non-
working married mothers relative to working mothers and single mothers).
     22 Kamerman & Kahn, supra note 17, at 205 (citing Finland and Sweden as articulat-

ing gender equity and encouraging women’s workforce participation as express objectives
of family beneªt policies).
     23 Id. at 214–16 (discussing various economic support initiatives in European countries

motivated at least partly by this objective).
     24
        See supra note 20 and accompanying text.
2005]                 A Defense of State Paid Family Leave                                  7

do receive some federal subsidies ªnanced through general revenues in
the form of tax credits. The Child and Dependent Care Tax Credit allows
working families with children to deduct a percentage of childcare expenses
from federal taxes owed.25 The Earned Income Tax Credit (“EITC”), a
refundable tax credit for low-income working families, gives more generous
credits to families with children.26 The Child Tax Credit allows taxpayers
to apply a credit against their tax amount on a per child basis.27 None of
these federal programs speciªcally addresses wage replacement for workers
who themselves take leaves to care for family members.
     State legislators have made some progress toward provision of wage
replacement for family caregiving, although here, too, coverage is lim-
ited. Most workers who receive any wage replacement during a family-
related leave from work do so by patching together maternity leave, sick
leave, and paid vacation beneªts.28 Just over half of all workers in private
industry have sick leave, i.e., accrued time off that an employee is enti-
tled to in the event of a nonwork-related personal illness.29 Sick leave
itself provides partial wage replacement for the period of actual incapaci-
tation due to childbirth, or in extraordinary cases, more extended preg-
nancy-related illness. Roughly one-third of workers in private industry have
paid short-term disability insurance.30 In a handful of states, these beneªts
are mandated through a TDI program.31 These beneªts are not, however,
directed toward any broader parental leave or family illness leave objectives

      25 The credit is a percentage, based on adjusted gross income, of the amount of work-

related childcare and dependent-care expenses an individual paid to a care provider to work or
look for work. See I.R.S. Pub. No. 503 (Dec. 11, 2003). There is a ceiling of $3,000 for
one qualifying dependent and $6,000 for two or more qualifying dependents. Id.
      26 The EITC reduces the amount of federal tax owed by workers and can result in a tax

refund for individuals if the credit exceeds the amount of taxes owed. See I.R.S. Pub. No.
596 (Dec. 31, 2003). Income and family size determine eligibility for the EITC. See id.
      27 26 U.S.C. § 24 (2004). The amount of the credit, unlike the EITC, cannot exceed the

amount of tax owed, and thus can result only in a lowered tax and not a refund.
      28 U.S. Census Bureau, Maternity Leave and Employment Patterns: 1961–

1995, at 12 tbl.G (2001) (indicating for leave taken after birth, 24.7% through maternity
leave beneªts, 8.8% through sick leave, 9.7% through vacation leave, and 1.0% through “other
paid leave”).
      29 Fifty-nine percent of employees in private establishments are covered by paid sick

leave plans. Bureau of Labor Statistics, U.S. Dep’t of Labor, USDL 04-53, Employee
Beneªts in Private Industry in the United States, March 2004, at 12 tbl.8 (2004)
[hereinafter Employee Beneªts], available at http://www.bls.gov/ncs/ebs/sp/ebsm0002.
pdf (last visited Nov. 18, 2004).
      30 In 2004, 39% of workers in private industry were covered by short-term disability

beneªts. Id. at 16 tbl.12. Short-term disability beneªts are an insurance beneªt funded by
contributions by the employer or employee (or both). Workers may combine sick leave and
short-term disability beneªts to increase the duration or amount of coverage.
      31 Five states (California, Hawaii, New Jersey, New York, and Rhode Island) and Puerto

Rico currently have TDI programs. Employment & Training Admin., U.S. Dep’t of
Labor, Temporary Disability Insurance 8-1 (2004) [hereinafter Temporary Disability],
available at http://workforcesecurity.doleta.gov/unemploy/uilawcompar/2004/temporary.pdf
(last visited Nov. 18, 2004). Of workers in private industry covered by short-term disability
beneªts in 2004, 16% (about 6% of all workers) were given the beneªts because of state
mandates or public provision. Employee Beneªts, supra note 29, at 22 tbl. 18.
8                        Harvard Journal of Law & Gender                           [Vol. 28

such as extended leaves by birth mothers who have medically recuper-
ated from childbirth for purposes of bonding with their newborns, leaves
by fathers or adoptive parents, or leaves to care for sick family members.
Some states have passed laws requiring employers that have sick leave
plans to permit employees to use their accrued beneªts to cover work ab-
sences to care for sick family members.32 Because most of these statutes
only apply to public employees, and also because many private employ-
ees are not actually covered by sick leave plans, their impact is limited.
Two states provide low-income workers some wage replacement to care
for newborns or newly adopted children.33 Also, a few states interpret
their unemployment insurance laws to allow workers forced to quit by rea-
son of pregnancy or work-family conºict to collect UI beneªts, but for
the most part, such work interruptions are considered voluntarily quits and
foreclose beneªts.34 With the exception of California, which I will dis-
cuss in more detail below, there are no public insurance programs compa-
rable to UI or TDI that are designed to provide workers with wage replace-
ment for family-care needs beyond pure maternity-related personal dis-
ability.
     In his ªnal days in ofªce, President Clinton directed the Department
of Labor to develop regulations that would allow states to use UI trust fund
monies to provide partial wage replacement to parents following the birth
or adoption of a child.35 The resulting regulation, Birth and Adoption Unem-
ployment Compensation (“BAA-UC”), enabled states to amend their unem-
ployment compensation statutes to exempt workers who voluntarily leave
a job for birth- or adoption-related reasons from the usual requirement



     32 States that permit extension of these rights to private employees are California, Cal.

Lab. Code § 233 (2004); Connecticut, Conn. Gen. Stat. § 31-51II (2003) (amended
2004); District of Columbia, D.C. Code Ann. § 32-502 (2004); Hawaii, Haw. Rev. Stat.
§ 398-3 (2003); Minnesota, Minn. Stat. § 181.9413 (2003) (applicable only for ill chil-
dren); Oregon, Or. Rev. Stat. §§ 659A.150–695A.186 (2003) (applies to use of paid vaca-
tion leave); Vermont, 21 Vt. Code Ann. § 472 (2004); Washington, Wash. Rev. Code
§ 49.12.270 (2004); Wisconsin, Wis. Stat. § 103.10 (2003).
     33 Missouri Early Childhood Development Fund, Mo. Ann. Stat. § 313.835 (1992);

Montana At-Home Infant Care Program, Mont. Code Ann. § 52-2-710 (2003).
     34 See Martin H. Malin, Unemployment Compensation in a Time of Increasing Work-

Family Conºicts, 29 U. Mich. J.L. Reform 131, 142 n.48 (1996) (reviewing decisions in
several jurisdictions ªnding work-family conºict to be cause for quitting); Deborah Maran-
ville, Changing Economy, Changing Lives: Unemployment Insurance and the Contingent
Workforce, 4 B.U. Pub. Int. L.J. 291, 317 n.107 (1995) (same); Mark R. Brown, A Case
for Pregnancy-Based Unemployment Insurance, 29 U. Mich. J.L. Reform 41, 44–55 (1996)
(describing how state unemployment compensation law treats maternity leave); David S.
Rosettenstein, Unemployment Beneªts and Family Policy in the United States, 20 Fam.
L.Q. 393, 398–403 (1986) (asserting that many state unemployment compensation law
systems prioritize the market over family).
     35 Press Release, The White House, Ofªce of the Press Secretary, President Clinton

Announces Actions to Offer Paid Leave to America’s Working Families (June 9, 2000),
available at http://clinton5.nara.gov/WH/New/html/20000612_1.html (last visited Nov. 18,
2004).
2005]                A Defense of State Paid Family Leave                               9

that they be “able and available” for work to collect beneªts.36 Two years
later, the Bush administration repealed the BAA-UC,37 citing, inter alia,
concerns about the solvency of state UI trust funds.38 As mentioned, how-
ever, several states are still pursuing legislative options for expanding or
creating social insurance programs to provide paid family leave.39
      Private establishments offer some coverage through voluntary pro-
grams, but it, too, is quite limited. A national survey of private establish-
ments (FLMA-covered and noncovered) in 2000 found that fewer than
half (42%) provided fully or partially paid maternity leave beneªts cover-
ing the period of physical incapacitation and recuperation immediately
surrounding childbirth.40 Still smaller percentages reported offering workers
fully or partially paid leaves of absence to care for a newborn (31%) or
seriously ill family member (31.8%).41 Eligibility limitations are likely to
mean that actual recipiency of paid beneªts is lower than the prevalence
of employer policies. A 1996 survey by the U.S. Census Bureau found that
only about one-third of women who worked during pregnancy actually
received pay during their post-birth work absence.42 Furthermore, the paid
beneªts that are available may be quite limited in duration. A study by
economists Jacob Klerman and Arleen Leibowitz found that although a
relatively high proportion of working women who took time off work fol-
lowing the birth of a child received paid beneªts in the immediate post-
birth period, recipiency sharply declined thereafter.43




    36  Birth and Adoption Unemployment Compensation, 20 C.F.R. § 604 (2000).
    37  Unemployment Compensation–Trust Fund Integrity Rule: Birth and Adoption Un-
employment Compensation; Removal of Regulations, 68 Fed. Reg. 58,540 (Oct. 9, 2003).
     38 Unemployment Compensation–Trust Fund Integrity Rule: Birth and Adoption Un-

employment Compensation; Removal of Regulations, 67 Fed. Reg. 233 (proposed Dec. 4,
2002).
     39 E.g., H.B. 25, 2003 Leg., 22d Sess. (Haw. 2003) and S.B. 772, 2003 Leg., 22d Sess.

(Haw. 2003); S.B. 778, 2003 Leg., 22d Sess. (Haw. 2003); H.P. 567, 2003 Leg., 121st Sess.
(Me. 2003); S.P. 389, 2003 Leg., 121st Sess. (Me. 2003); A.B. 173, 2004 Leg., 211th Sess.
(N.J. 2004); H.B. 2399, 2004 Leg., 58th Sess. (Wash. 2004.), S.B. 6272, 2004 Leg., 58th
Sess. (Wash. 2004).
     40 Balancing the Needs, supra note 5, at tbl.5.6. A further 15.4% of establishments

stated that pay would depend on circumstances. Id.
     41 Id. Again, some establishments stated that pay would depend on circumstances (18%

for a newborn and 23% for a family member).
     42 U.S. Census Bureau, supra note 28, at 12 tbl.G.
     43 Jacob Alex Klerman & Arleen Leibowitz, The Work-Employment Distinction Among

New Mothers, 29 J. Hum. Resources 277, 299 tbl.A3 (1994) (showing that 41% of em-
ployed mothers were on paid leave when their infant was one month old, and 16% were on
paid leave when their infant was two months old, and by the third month, less than 5%
were receiving any form of wage replacement). These data are fairly old (1986–88), but are
consistent with the continuing practice of making beneªts linked to physical recuperation
from birth more available than more extended family leave beneªts.
10                      Harvard Journal of Law & Gender                          [Vol. 28

        B. Why Private Markets Fail To Provide Paid Family Leave

     An important question in thinking about any labor market interven-
tion is why the practice in question does not arise in private markets. Many
beneªts in the United States are linked to employment, with the employer
either self-insuring or acting as an intermediary between the workers and
an insurance provider. In theory, then, workers will bargain for beneªts they
desire as part of their employment compensation packages.
     Perhaps, then, the limited provision of insurance to cover leaves of
absence to deal with family illness or caring for infant children reºects
the fact that workers do not want such insurance. If paid leave were impor-
tant to them, those workers who expect to have children or experience other
family-related work conºicts could elect to substitute some wage compensa-
tion now for a promise of guaranteed wage replacement at a later time
when work interruption occurs. However, there are a number of reasons why
we might fail to observe this kind of bargaining even assuming family
leave is valuable to some workers.
     First, as readers of the literature on labor market regulation more gener-
ally will recognize, scholars have identiªed a number of information-based
bargaining failures that might explain private market failures to provide
workplace beneªts that workers desire.44
     Some explanations turn on mistakes workers are apt to make about
their own interests. Workers might, for example, miscalculate their down-
stream needs because of poor information or because cognitive bias leads
them to systematically underestimate future needs.45 They might also sys-
tematically discount the value of future, as opposed to present, income, a
phenomenon known to economists as “hyperbolic discounting.”46 In addi-
tion, ex ante entitlements such as wealth might affect the value workers
place on receiving wage versus non-wage compensation. Low-wage work-
ers, for whom cash is in shorter supply, may believe that higher wages in
the present are preferable to the promise of beneªts at some future time,
although they may later regret having made this tradeoff.47 In other cases,


     44 For general reviews of the extensive literature on bargaining failures in labor mar-

kets, see Richard Edwards, Rights at Work: Employment Relations in the Post-
Union Era 42–76 (1993); Paul Weiler, Governing the Workplace: The Future of
Labor and Employment Law 74–78 (1990).
     45 See Christine Jolls, Cass R. Sunstein & Richard Thaler, A Behavioral Approach to

Law and Economics, 50 Stan. L. Rev. 1471, 1476–79 (1998) (summarizing research ªndings
on the effect of various kinds of informational and cognitive failures that might lead to
poor labor market decisions by workers); Deborah Weiss, Paternalistic Pension Policy:
Psychological Evidence and Economic Theory, 58 U. Chi. L. Rev. 1275, 1297–1311
(1991) (discussing savings failures caused by “myopia” (the irrational tendency to favor
present over future consumption), time-inconsistent preferences, and impulsiveness).
     46 See generally David Laibson, Golden Eggs and Hyperbolic Discounting, 112 Q. J.

Econ. 443 (1997).
     47 Joseph Bankman, Tax Policy and Retirement Income: Are Pension Plan Antidis-

crimination Provisions Desirable?, 55 U. Chi. L. Rev. 790, 808–09 (1988) (suggesting
2005]                 A Defense of State Paid Family Leave                                 11

workers who desire family leave beneªts might not demand them for fear
that they will signal to their prospective employer that they are likely to
need them (and thus might be less reliable than other workers). Conversely,
a desire to avoid attracting a disproportionate share of workers likely to
need leaves may lead employers to resist offering beneªts that workers
would be willing to purchase.48
     Collective action problems can also prevent workers from bargaining
toward the outcome they desire: no one worker may be willing to incur the
upfront costs of pressing for a new kind of beneªt, and no good mechanism
may be in place to help workers coordinate to overcome this problem.
The fact that some workers have no need for family leave policies may make
collective action even more difªcult. Unions, although they may serve a co-
ordination role, only represent a small percentage of American workers.49
     In addition to these general labor market bargaining failures, family
leave beneªts may pose special challenges for private insurance markets.
The two kinds of family leave beneªts, family illness leave and parental
leave, each present a distinct set of problems.
     First consider family illness leave. When a worker’s family member
unexpectedly becomes ill, and the worker takes time off to provide care, the
resulting lost wages may impose signiªcant hardship. Protection against
the risk of unexpected changes in fortune is traditionally the province of
insurance, the social device that pools risks.50 Insurance forms the back-
bone of healthcare ªnancing in the United States.51 For example, about 60%
of Americans have private health insurance, most of them through employer
plans.52 Most employers that offer health insurance also offer wage re-

that rank-and-ªle employees may place less value on pension beneªts, at the expense of
salary, than more highly compensated employees).
     48 David Charny, The Employee Welfare State in Transition, 74 Tex. L. Rev. 1601,

1618 (1996) (suggesting that һrms may offer less extensive insurance coverage than
workers would optimally purchase, out of a fear that more generous [coverage] would attract
high-risk workers”).
     49 See Bureau of Labor Statistics, U.S. Dep’t of Labor, Union Members in

2003, at 1 (2004) (citing that 12.9% of American workers in 2003 were represented by
unions). Note that a 2004 survey of California employees found that unionized employees
were 3.6 times more likely than nonunion employees to provide family and medical be-
yond those required by law. See Ruth Milkman & Eileen Appelbaum, Paid Family Leave
in California: New Research Findings 6 (UCLA Institute of Industrial Relations, Re-
search Brief, 2004).
     50 Robert Riegel & Jerome S. Miller, Insurance Principles and Practices 26

(4th ed. 1959) (“Insurance is a social device whereby the uncertain risks of individuals
may be combined in a group and thus made more certain, small periodic contributions by
the individuals providing a fund out of which those who suffer losses may be reimbursed.”).
     51 David M. Cutler & Richard J. Zeckhauser, The Anatomy of Health Insurance, in 1A

Handbook of Health Economics 563, 566 (Anthony J. Culyer & Joseph P. Newhouse
eds., 2000).
     52 Id. at 570 tbl.1 (showing 56% through employer-sponsored plans, 6% through direct

purchase). Roughly 25% of the population is insured through the public sector (e.g., Medi-
care and Medicaid), and 16% remains uninsured. Id. The predominance of employer spon-
sorship in the private health care market is likely the result of two factors: (1) the tax-free
nature of employer-provided insurance contributions, as compared with wages, which are
12                      Harvard Journal of Law & Gender                           [Vol. 28

placement beneªts for personal illness (sick leave).53 In addition, almost
all employer plans offer family health insurance coverage.54 One might
think that a plausible extension of family coverage would be insurance
covering wage loss in the event of job interruptions caused by family ill-
ness, analogous to personal sick leave. Like sick leave, paid family illness
leave would cover the opportunity costs (lost wages) of illness rather than
the direct costs (medical fees, hospital bills). Yet paid family illness leave
is uncommon. What makes it so scarce in private markets?
      Insurance in private markets uses an actuarial mechanism. The
amount an individual is required to contribute to the pool, called the pre-
mium, varies with the level of risk insured. The cost of the premium cannot
exceed the monetary equivalent of the expected disutility of self-insuring.55
If a person’s risk of experiencing a hazard is 100%, she will be unable to
purchase private insurance. The market (or lack thereof) will simply
reºect the fact that the premium for purchasing coverage would be the same
as (or even higher than, given transaction costs) the cost of saving for the
inevitable occurrence. An example of the kind of risk that might be ex-
cluded from private insurance because the likelihood of its occurrence
approaches certainty is a preexisting condition such as a congenital ill-
ness.
      From the outset, insurance against family illness-related work leaves
would need to contemplate a much higher likelihood of hazard than cov-
erage of personal illness leaves. Instead of merely contemplating the risk
of a particular worker suffering a debilitating illness or injury, the pool
would need to consider the probability of a debilitating illness or injury
on the part of the worker, her partner, her parents, or her children. In ad-
dition, the average risk level of each individual whose illness might trig-
ger eligibility for beneªts would likely increase. On the whole, adult work-
ers are probably more robust and self-sufªcient than the children, elders,
and disabled adults who might depend on them. Thus the aggregate level
of risk in the pool would increase as the scope of coverage broadened, mak-
ing insurance against these risks more costly to supply.
      In addition, information problems would pose formidable barriers to
private family illness insurance. Information problems can often cause fail-
ures in insurance markets. If insurance companies had full information about
every applicant, they would be able to charge each person a premium that

taxed, and (2) the likelihood of pooling economies associated with group insurance pur-
chases. Jonathan Gruber, Health Insurance and the Labor Market, in 1A Handbook of
Health Economics, supra note 51, at 645, 651.
     53 See Gruber, supra note 52, at 653 tbl.2 (showing that 71.1% of ªrms that offered

health insurance in 1993 also offered short-term disability beneªts).
     54 See id. (showing that 91% of employers that offered health insurance in 1993 also

offered family coverage).
     55 A competitive insurance premium would be B = p L + T, where p is the probability
                                                     i    i               i
of the insured event occurring, L is the magnitude of the insured loss, and T represents the
transaction costs.
2005]                 A Defense of State Paid Family Leave                               13

perfectly reºected the risk of her need for indemniªcation. “Low-risk”
individuals would pay low premiums, and “high-risk” individuals would
pay high premiums. As mentioned above, some individuals—those cer-
tain to experience the hazard—would be priced out of the market entirely. In
many areas of insurance, insurers do a good enough job of estimating the
risks of applicants that they are able to offer insurance at a price that is
both affordable to purchasers and proªtable to insurers.
      Where good information is not available to insurers, however, they
may stay out of the market.56 One informational barrier to insurance is the
problem of adverse selection. If the insurer is unable to distinguish be-
tween high- and low-risk individuals (perhaps because risk is not obvious
to outside observers, and high-risk individuals have an incentive not to
disclose information likely to increase their insurance rates), it will be
unable to price insurance efªciently. Averaging the price of premiums is
no solution, because low-risk individuals—assuming they know that they
are low-risk individuals—are likely to ªnd the “averaged” premiums higher
than they are willing to pay and drop out of the market.57 As prices charged
to those left behind rise, and more participants drop out with each suc-
cessive increase in premiums, the market unravels.
      When a worker is herself incapacitated due to illness or injury, the
decision of whether to work is, in theory, relatively nondiscretionary. The
question for the insurer is authentication of the claim of inability to work,
usually done through a doctor’s veriªcation of the existence and severity
of the illness. When a worker’s family member becomes ill or disabled
(even assuming adequate veriªcation), the decision as to whether the in-
sured worker must absent herself from work to care for the ill family mem-
ber may involve a weighing of personal priorities and alternative options.
This heightens the potential for adverse selection problems due to the in-
surer’s inability to anticipate what kinds of decisions different workers are
likely to make.
      Another likely barrier to family illness leave is moral hazard. One
who must bear the full costs of her behavior will take precautions to reduce
risk and stem losses. A person with insurance pays only the cost of pre-
miums, rather than the full cost of loss. Because losses are cheaper for
someone who has insurance, she may take fewer precautions. This phe-
nomenon, known as moral hazard,58 would not be a problem if insurers

     56 My discussion here of adverse selection, moral hazard, and the potential role for the

state in providing insurance relies extensively on Nicholas Barr, Economic Theory and the
Welfare State: A Survey and Interpretation, 30 J. Econ. Literature 741, 750–53 (1992).
For more extensive citations, see Gillian Lester, Unemployment Insurance and Wealth Redis-
tribution, 49 UCLA L. Rev. 335, 360–65 (2001) [hereinafter Lester, Wealth Redistribu-
tion].
     57 Offering two different premiums also fails for similar reasons. Barr, supra note 56,

at 751–52.
     58 Economists distinguish further between ex ante moral hazard (referring to the situa-

tion before the hazard) and ex post moral hazard (referring to the avoidance of further
14                     Harvard Journal of Law & Gender                         [Vol. 28

could fully observe people’s levels of precaution and price premiums or
restrict coverage accordingly. However, insurers often lack full informa-
tion about individual precaution, and so they use copayments, deductibles,
partial indemniªcation, or other devices to make insured individuals bear
some of the costs of their own carelessness. If an insured event is very psy-
chologically or physically costly (e.g., serious personal injury) the in-
sured person’s fear of harm may naturally temper moral hazard. But some-
times a person prefers that an insured event occur; the only barrier is that
without insurance, it costs more than she is willing to pay. For example,
elective cosmetic surgery is often excluded from health insurance poli-
cies for this reason.59 In such cases, problems of moral hazard will usu-
ally make insurance impossible.
      If wage replacement were available for family illness leave, the in-
sured’s incentive to avoid the insured event—in this case lost wages due
to job interruption—would decline. For some workers, wage replacement
would have no effect on their decisions. Sometimes, the decision to take
a family illness leave as a practical matter is just as nondiscretionary as
in the run of personal-illness cases: the worker may prefer not to inter-
rupt her employment, but great potential hardship to a family member
lies in the balance, with no realistic alternatives available. But in other cases,
wage replacement would tip the balance for workers who wanted to take
leave but felt they could not afford it. A private insurer may decide, for
this reason, that paid family leave, like elective cosmetic surgery, pre-
sents too great a moral hazard problem. Paid family leave insurance may
well trigger some inefªcient decisions. For example, a worker whose pro-
ductive value is $300 per day might decide to take time off of work to
care for an ill family member instead of hiring an external caregiver at a
cost of $100 per day. Insurance against wage replacement would make it
easier to do this. As I will discuss in detail later, there may be good rea-
sons, beyond efªcient allocation of caregiver human capital, for making
paid leave available even to this worker. For the moment, however, my point
is that a private insurer might worry about inefªcient overutilization of
beneªts and decline to enter this market.
      Next, consider paid parental leave. From the outset, actuarial insur-
ance (i.e., insurance against uncertain risk) seems unsuited to wage losses
associated with parental leaves.60 Squeezing the event of bearing or adopting

losses after the hazard). See Peter Zweifel & Willard G. Manning, Moral Hazard and Con-
sumer Incentives in Health Care, in 1A Handbook of Health Economics, supra note 51,
at 409, 413. Ex post moral hazard is at issue in the context of job leaves.
     59 Barr, supra note 56, at 780.
     60 Of course, some pregnancies are unplanned (and some unplanned pregnancies are

unwanted). In 1994, 49% of pregnancies in the United States were unintended, meaning that
they ended in abortion, the woman had wanted no children at that time, or she had wanted
no more children ever. Stanley K. Henshaw, Unintended Pregnancy in the United States,
30 Fam. Plan. Persp. 24, 26 tbl.1 (1998). Of these unintended pregnancies, 54% ended in
abortion. Id.
2005]                 A Defense of State Paid Family Leave                              15

a child into actuarial parameters encounters overwhelming problems of
moral hazard. Whereas illness of a worker is usually unexpected and psy-
chologically costly, as is a worker’s need to miss work to care for family
members, the birth or adoption of a child is normally anticipated and de-
sired, even if the timing cannot be planned with precision.61 And, unlike
an illness, which can potentially strike any worker, bearing or adopting a
child is certain not to happen for some workers. This stands in contrast to
sick leave, in which even those who will never use beneªts do not know
this from the start, and thus will probably get some value from feeling
protected in the event of possible bad luck.
     With parental leave, the kind of need in question is not so much in-
demniªcation against unexpected, unwanted events as it is a need to en-
sure that adequate funds are available for workers to manage predictable,
and often chosen, family demands that interrupt work. In other words, there
is a need for redistribution of a worker’s own income across the life cy-
cle. One question is why workers do not simply borrow money or save for
predictable interruptions in wage income due to the birth of a child. I have
already discussed various cognitive failures that may lead workers to un-
derestimate their future needs. In addition, however, even assuming per-
fectly rational planning, workers with families may have a difªcult time
accumulating savings to protect themselves against this kind of loss. For
most families, the arrival of children corresponds to the parents’ early
career years, when wages are relatively low, and the opportunity to set
aside surplus earnings for savings is more constrained. These workers, or
other workers who have very low incomes throughout their working years,
may ªnd that the cost of daily life leaves them without savings despite a
desire to save. In fact, Americans are very poor savers.62 Recall that ªnancial
constraints prevent some workers from taking leaves, and workers who
do take leaves have higher incomes and salaries than those who do not.63
This suggests, although it does not prove, that for workers who do not
take leave, their reasons may go beyond simply having other ªxed costs
like a mortgage or a college tuition debt resulting from other projects that
enhance their quality of life and opportunity. Finally, capital markets are




     61 See, e.g., Meryl Frank, Costs, Financing, and Implementation Mechanisms of Paren-

tal Leave Policies, in The Parental Leave Crisis: Toward a National Policy 315, 320
(Edward F. Zigler & Meryl Frank eds., 1988) [hereinafter The Parental Leave Crisis]
(stating that a program of salary replacement for infant care does not meet the deªnition of
insurance because pregnancy is typically a planned event with a known time and duration,
and thus there is no “risk” against which to insure).
     62 See Eric Engen & Jonathan Gruber, Unemployment Insurance and Precau-

tionary Savings (Nat’l Bureau of Econ. Research, Working Paper No. 5252, 1995)
(showing that the median twenty-ªve- to sixty-two-year-old worker has gross ªnancial assets
equal to less than three weeks’ income).
     63
        See supra note 7.
16                      Harvard Journal of Law & Gender                          [Vol. 28

not widely accessible to workers who wish to borrow against the promise
of future earnings.64
     In sum, private markets generally fail to provide family leave insur-
ance for several reasons—some based on information or cognitive failures
that may arise in all forms of bargaining between workers and employers,
and some based on distinct challenges associated with pooling risks efª-
ciently for this kind of insurance. It is possible that even if we could cor-
rect all market failures, we would conclude that private markets do un-
dersupply paid leave. This would be so, for example, if paid leave beneªts
advance goals aside from aggregate efªciency that we nonetheless deem
to be socially valuable.

                        C. The Role of State Intervention

     Where private markets fail to supply needed economic security—
whether due to employment-related bargaining failures, private savings and
capital market failures, insurance market failures, or other reasons—the
state can play a role, either through public provision or by mandating some
form of private provision. However, state intervention is not justiªed simply
because private markets do not provide particular forms of economic se-
curity and state provision (or facilitation) is feasible. The normative de-
sirability of compulsory provision as well as the particular form of com-
pulsory provision must be defended.
     Assuming private markets produce less paid leave than society deems
desirable, the state can increase its prevalence in a range of ways. The
choice of mechanism depends on our reasons for valuing the beneªt. If
we want to encourage some kind of behavior because we believe people
will otherwise make poor decisions (perhaps because of informational or
cognitive failures), the state can create incentives simply to correct “er-
ror” and encourage better personal decisions. We might, on the other hand,
want to encourage behavior that has value beyond the beneªt to those
directly affected. In such a case, we may wish to ªnance the program in a
way that deliberately spreads its costs across society. The redistributional
level of a program turns both on who receives beneªts and how revenues
are generated to pay for beneªts.
     To illustrate, we may decide that it is valuable for individuals to have
funds to pay for out-of-pocket medical expenses during the course of the
year, but ªnd that individuals are generally poor savers and tend not to
save for this purpose. One way the state could ameliorate this problem is
to offer tax incentives to individuals who save. Federal Health Savings



     64 See Stephen P. Zeldes, Consumption and Liquidity Constraints: An Empirical Inves-

tigation, 97 J. Pol. Econ. 305, 305 (1989) (ªnding that a signiªcant portion of the popula-
tion is constrained in the ability to borrow against the promise of future labor income).
2005]                  A Defense of State Paid Family Leave                                  17

Accounts are a version of this.65 Here, the beneªt is available only to those
who make contributions and only in the amount they contributed. The tax
incentive essentially encourages self-insurance. Redistribution occurs, but is
limited to the tax savings enjoyed by those who participate in the program.
      By contrast, the government may decide to create a program that is
more aggressively redistributive. For example, the government may de-
cide to make health insurance available to those who would be excluded
from private markets on the basis of their health characteristics. As ex-
plained in the foregoing discussion, people with very high-risk character-
istics will generally be unable to purchase private insurance; they will need
to self-insure. By mandating across-the-board participation in an insur-
ance pool, the government could force broader spreading of risks than we
would see in a well-functioning private insurance market.66 The state
might do this believing that everyone materially beneªts if economic secu-
rity is widely distributed, or it might do so as a way to increase equality
or social solidarity.67 Medicare, which provides health insurance to elderly
and disabled workers but is ªnanced by a tax on all workers, depends on
signiªcant redistribution between low-risk and high-risk individuals.68
      Although I have illustrated this with only two examples, a wide spec-
trum of redistribution is possible. The point is that any proposal for state-
facilitated paid family leave requires an analysis of what motivates the

     65 26 U.S.C.S. § 223 (2004). The program allows an individual to make a tax deduction

on an amount placed into a trust that is set up to allow her to pay for medical expenses of
named beneªciaries. The maximum amount of the deduction is based on the yearly sum of
monthly payments placed into the account, with monthly payments limited according to
the type and level of deductible of their health insurance plan.
     66 Recall that in private actuarial insurance, individuals voluntarily pool their resources.

When a member of the pool suffers a covered loss and therefore draws on the pool, there is
redistribution between members of the pool. This constrained risk-spreading within a risk-
adjusted (actuarial) insurance pool is part of a bargain between participants who have compa-
rable risk characteristics or have paid different premiums based on their risk characteristics.
     67 A number of scholars have suggested that morality requires the state to provide

health insurance to “high-risk” people who would be excluded from private insurance. See,
e.g., Norman Daniels, Health-Care Needs and Distributive Justice, 10 Phil. & Pub. Aff.
146, 165, 175 (1981) (applying John Rawls’s theory of justice to the health care context
and concluding that health care institutions should be included on the list of basic institu-
tions regulated by a fair equality of opportunity principle, and that no persons should face
barriers, ªnancial or otherwise, to initial access to the health care system); Donald W.
Light, The Practice and Ethics of Risk-Rated Health Insurance, 267 JAMA 2503, 2507–08
(1992) (arguing that actuarial fairness in health insurance is morally unfair because it re-
duces access to life opportunities and increases suffering for those disadvantaged by risk,
pain, and illness); Deborah A. Stone, At Risk in the Welfare State, 56 Soc. Res. 591, 623–
33 (1989) (arguing that the use of health-risk classiªcation for rationalizing health care
treats disadvantaged individuals as undeserving and saddles those who are already in need
with additional burdens). On the potential for universalistic social insurance schemes to
increase social solidarity by incorporating both lower and middle classes, see GØsta Esp-
ing-Andersen, The Three Worlds of Welfare Capitalism 31–33 (1990).
     68 Kyle Logue & Ronen Avraham, Redistributing Optimally: Of Tax Rules, Legal

Rules, and Insurance, 56 Tax L. Rev. 157, 226–27 (2003) (commenting on the high degree
of cross-subsidization between the healthy and the sick in Medicare due to the fact that it is
funded by a payroll tax that does not distinguish between individuals based on health risks).
18                       Harvard Journal of Law & Gender                            [Vol. 28

proposal. The underlying normative justiªcations for intervention will have
important implications for program design, in particular, how broadly the
costs of the program should be shared. With that in mind, let us turn to
the question of why a more equal distribution between men and women
of family care obligations and participation in paid employment is impor-
tant and how paid leave can help to achieve this goal.

     III. Women, Family Care, and Labor Market Participation

     In this Part, I analyze the economic evidence that there is gender
inequality in existing work and family arrangements. Second, I analyze why
this inequality matters. In part, I argue that equalization of burdens and
beneªts at the work/family nexus is consistent with broader social equal-
ity ideals, and as such is something that all members of society have an
interest in achieving. I also argue, based on psychological and sociologi-
cal evidence, that participation in paid employment can increase women’s
capacity to ºourish.

     A. Bonding, Family Health, and Children as a Collective Concern

     A number of scholars have defended family leave on the basis of its
potential to improve child and elder welfare. A sustained period of paren-
tal care for a newborn infant would give working parents an opportunity to
care for and bond with infant children. This bonding between parent and
child may have distinctive beneªts for both—a sense of personal achieve-
ment and intimacy on the part of the parent, and improved health on the
part of the infant.69 In addition, parents who are able to minimize ªnancial
strain while caring for infants may be better able to reduce family stress
and create a safe, well-functioning household.70 In the case of sick chil-
dren, studies ªnd that the presence of a parent lessens the severity of
symptoms and speeds recovery.71 Paid leave policies are themselves posi-

     69 See Doriane Lambert Coleman, Fixing Columbine: The Challenge to Ameri-

can Liberalism 48–50 (2002) (discussing attachment theory and its detractors); T. Berry
Brazelton, Issues for Working Parents, in The Parental Leave Crisis, supra note 61, at
36, 39 (claiming that there is likely to be a signiªcant improvement in a mother’s feelings
of personal achievement and intimacy with her baby if she is able to remain home from work
for at least three months after birth); C. R. Winegarden & Paula M. Bracy, Demographic
Consequences of Maternal-Leave Programs in Industrial Countries: Evidence from Fixed-
Effects Models, 61 S. Econ. J. 1020 (1995) (studying the effect of increased generosity in paid
maternal leave policies in seventeen OECD countries between 1959 and 1989 and ªnding an
inverse correlation between duration of paid maternity leave available and infant mortality).
     70 Robin Harwood, Parental Stress and the Young Infant’s Needs, in The Parental

Leave Crisis, supra note 61, at 55 (reviewing studies ªnding that high levels of parental
stress may hinder parents’ ability to be emotionally responsive to their infants, suggesting
that among the most vulnerable are low-income or single working parents, and arguing that
a curtailed postpartum work schedule without ªnancial loss may signiªcantly reduce the
stressfulness of this time period).
     71
        Comm. on Hosp. Care, Am. Acad. of Pediatrics, Family-Centered Care and the Phy-
2005]                 A Defense of State Paid Family Leave                                 19

tively correlated with better child health outcomes, perhaps because they
increase working parents’ ability to care personally for their children, and
this care is better than the alternatives relied on in its absence.72 At a mini-
mum, then, these arguments suggest that individuals who lack the savings to
take a leave from work to care for their infants or sick children might be
better off if the state intervened to encourage better ªnancial planning.
      Some defend state involvement in mandated paid leave policies on
the basis that the beneªts to children have broader value to society. They
argue that caring for infants and children is a collective concern because
it is an investment in the future of society: healthy, well-adjusted children
will eventually become good citizens and workers.73 Some emphasize bur-
dens on caregivers more generally, arguing that society owes a debt to those
who perform these society-preserving labors.74 They make the liberal claim
that caregiving labor is a public responsibility because it ensures the
conditions for children’s development into autonomous adults.75

sician’s Role, 112 Pediatrics 691, 693 (2003) (reviewing medical literature showing that
family presence while children are sick, inter alia, decreases child anxiety, speeds recov-
ery, and reduces need for medication). See generally Jody Heymann, The Widening
Gap: Why America’s Working Families Are in Jeopardy and What Can Be Done
About It 57 (2000) (citing several studies ªnding a positive correlation between the pres-
ence of a parent and health outcomes of sick children).
     72 Christopher Ruhm, Parental Leave and Child Health, 19 J. Health Econ. 931, 933,

947 (2000) (ªnding rights to paid parental leave in sixteen European countries over twenty-
ªve–year period associated with substantial decreases in pediatric mortality, but no rela-
tionship between rights to unpaid leave and infant mortality).
     73 Nancy Dowd, Family Values and Valuing Family: A Blueprint for Family Leave, 30

Harv. J. on Legis. 335, 345–47 (1993) (defending public subsidy of family leave policies
on the grounds that children are a societal concern, beneªt, and investment in the future for
all of us); Marianne A. Ferber, Commentary on Chapter 5, in Gender and Family Issues
in the Workplace 162, 162 (Francine D. Blau & Ronald G. Ehrenberg eds., 1997) (argu-
ing that parental leaves are useful because children will do better when cared for in their
home for some months after birth, and such children will grow up to be productive workers
and good citizens); Young, supra note 1, at 158 (defending state-subsidized paid leave
policies on the basis that society has an economic interest in the well-being of its children).
But see Mary Anne Case, How High the Apple Pie? A Few Troubling Questions about
Where, Why, and How High the Burden of Care for Children Should be Shifted, 76 Chi.-
Kent L. Rev. 1753, 1775–78 (2001) (criticizing efforts to characterize children as “public
goods” to justify subsidies to parents); Katherine M. Franke, Theorizing Yes: An Essay on
Feminism, Law and Desire, 101 Colum. L. Rev. 181, 192–95 (2001) (criticizing arguments
that characterize children as the next generation of workers to justify subsidies). See gen-
erally Martha Albertson Fineman, Cracking the Foundational Myths: Independence, Auton-
omy, and Self-Sufªciency, 8 Am. U.J. Gender Soc. Pol’y & L. 13, 19 (2000) (arguing that
caretaking labor provides for citizens, voters, consumers, students, and others who popu-
late state institutions, and because these institutions depend on such labor, it should be
compensated by the state).
     74 See Martha Fineman, The Neutered Mother, The Sexual Family and Other

Twentieth Century Tragedies 161–64 (1995) (arguing that society must recognize the
material costs and consequences of the uncompensated caretaking burden borne largely by
women); Fineman, supra note 73, at 18–19 (arguing that caretaking labor is an unrecog-
nized subsidy to all of society and as such should be compensated by the state).
     75 See, e.g., Linda McClain, Care as a Public Value: Linking Responsibility, Resources,

and Republicanism, 76 Chi.-Kent L. Rev. 1673, 1683 (2001) (arguing that care for chil-
dren, which fosters children’s capacity to live successful lives and be good citizens, is a
20                       Harvard Journal of Law & Gender                            [Vol. 28

     In the case of elders, a family caregiver can assist in many of the
tasks necessary for daily living, but can also be a source of emotional
support.76 The responsibility of giving personal care to disabled elders can
be extremely stressful for family members.77 Easing the ªnancial burden
of taking time away from work to provide care, and thus reducing some
of this stress, may improve the health of both the ill family member and
the worker who gives care. Improving the life conditions of disabled eld-
ers and their caregivers in this way also may redound to the beneªt of
society as a whole by reducing the social burden of health care.
     These arguments have some force, but the question remains why paid
leave policies, as opposed to other policies that might advance the same
ends, are desirable. Society might enjoy the beneªts listed above to an even
greater degree if social policy were designed to encourage single-earner
partnerships: a ªnancially supported adult staying in the home could de-
vote greater time than a worker to child and eldercare. Alternatively, we
might create a more comprehensive infrastructure for externally provided
caregiving, such as publicly provided daycare, or state-subsidized home
healthcare and eldercare provision. Paid leave, although it may facilitate
parent-child bonding and family health, may not, in fact, be the best method
for maximizing these goods. My defense of paid family leave instead
rests principally on other normative goals: increasing labor market equal-
ity and enhancing women’s autonomy.

                   B. Women, Caregiving, and the Family Gap

     Although estimates vary somewhat across studies, researchers consis-
tently ªnd that women spend signiªcantly more time than men engaged in
caregiving activities for children78 and the disabled elderly.79 Many women

public value and responsibility).
     76 Nadine Taub, From Parental Leaves to Nurturing Leaves, 13 N.Y.U. Rev. L. & Soc.

Change 381, 386–87 (1984-85) (arguing that families can provide not only the physical
care that would be available in an institutional setting, but also love, intimacy, affection,
and solidarity).
     77 Id. at 387; Lynn M. Martire & Mary Ann Parris Stephens, Juggling Parent Care and

Employment Responsibilities: The Dilemmas of Adult Daughter Caregivers in the Work-
force, 48 Sex Roles 167 (2003).
     78 See Joseph H. Pleck, Paternal Involvement by U.S. Residential Fathers: Levels, Sources,

and Consequences, in The Role of the Father in Child Development 66, 71 (Michael
E. Lamb ed., 3d ed. 1997) (meta-analysis of research ªndings during the mid-1980s and
early 1990s, estimating that fathers spend an average of 43.5% of the time that mothers
spend directly engaged in childcare activities, and that their accessibility, deªned as avail-
ability to the child in the absence of direct involvement, is on average 65.6% of the acces-
sibility of mothers). More recently, see Bureau of Labor Statistics, U.S. Dep’t of
Labor, USDL 04-1797, Time-Use Survey—First Results Announced by BLS, at tbl.7
(2004), available at http://www.bls.gov/tus (last visited Nov. 18, 2004) (reporting results of
2003 survey of American adults in households with children under eighteen years of age,
which found that men spent 46% of the time women did engaged in the primary activity of
caring for household children).
     79
        See Robyn I. Stone & Peter Kemper, Spouses and Children of Disabled Elders: How
2005]                 A Defense of State Paid Family Leave                                21

with caregiving obligations also hold paid employment.80 Even when they
are working full time, women still invest considerably more time than men
caring for their children and disabled parents.81 For many women, though,
family care obligations lead them to loosen their attachment to the work-
force. Women are more likely than men to interrupt work or quit their jobs
to care for children or parents.82 In addition, some women are deterred from
working at all because of these obligations.83 Decisions within the hetero-
sexual household to have the woman rather than the man decline or reduce
paid work to give care are partly the result of divergent opportunity costs
due to the gender wage gap.84 Yet this creates a self-reinforcing cycle of
gender inequality—the different labor market wages and opportunities of
men and women inºuence household decisions about allocating home ver-
sus market work, and these decisions in turn reinforce the wage gap.85

Large a Constituency for Long-Term Care Reform?, 67 Milbank Q. 485, 494 (Milbank
Mem’l Fund 1990) (1989) (“Among children who are primary caregivers [of disabled par-
ents], daughters outnumber sons by more than 3 to 1; among primary caregivers of more
disabled elders, the ratio is even higher—4 to 1.”); Robyn Stone et al., Caregivers of the
Frail Elderly, 27 Gerontologist 620, 620 (1987) [hereinafter Stone et al., Caregivers]
(estimating that daughters were twice as likely as sons to assume primary responsibility for
the frail elderly with no assistance).
     80 See, e.g., Stone et al., Caregivers, supra note 79, at 620 (ªnding that 44% of adult

daughter caregivers are employed).
     81 Organisation for Economic Co-Operation and Development, OECD Em-

ployment Outlook 137, 140 tbl.4.5 (2001) (reporting that U.S. fathers in 1995 spent an
average of thirty-three minutes per day, compared to sixty-two minutes per day for full-
time working mothers, engaging in childcare activities, deªned as feeding, dressing, changing,
bathing, and giving medication to children); see also Eleanor Palo Stoller, Parental Care-
giving by Adult Children, 45 J. Marriage & Fam. 851, 856 (1983) (being employed de-
creases the average level of sons’ assistance in parental caregiving by 22.9 hours per
month, but has no effect on levels of assistance by daughters, who might make up the time
by reducing leisure).
     82 Balancing the Needs, supra note 5, at tbl.A2-2.6 (showing that 26.7% of women

who took a leave from work in 2000, whether or not covered by the FMLA, used the leave
to care for children (not including leave taken to care for newborns or newly adopted chil-
dren) or elders, as compared with 21.4% of male leave-takers); Kristen Keith & Abigail
McWilliams, The Returns to Mobility and Job Search by Gender, 52 Indus. & Lab. Rel.
Rev. 460, 465 tbl.1 (1999) (estimating that 8.4% of women, compared with 3.8% of men,
quit their jobs for family-related reasons between 1979 and 1984); Stone et al., Caregivers,
supra note 79, at 622 tbl.3 (ªnding that of adult children with eldercare responsibilities,
11.6% of daughters, compared with 5% of sons quit their jobs as a result).
     83 Bureau of Labor Statistics, U.S. Dep’t of Labor, Persons Not in the Labor

Force by Desire and Availability for Work, Age, and Sex, Chart A-37 (2004), avail-
able at ftp://ftp.bls.gov/pub/suppl/empsit.cpseea37.txt (last visited Nov. 18, 2004) (show-
ing that among those who were not in the labor force in September 2004 and not available
to work but wanted a job, 44.96% of women but only 11.89% of men reported family re-
sponsibilities as the reason for not working).
     84 Christine A. Littleton, Does It Still Make Sense To Talk About Women?, 1 UCLA

Women’s L.J. 15, 36 (1991) (noting that the wage gap between men and women makes it
economically more rational for married women than their husbands to take an unpaid leave
from work to engage in caretaking).
     85 Joni Hersch & Leslie S. Stratton, Housework, Wages, and the Division of Housework

Time for Employed Spouses, 84 Am. Econ. Rev. 120, 124 (1994) (identifying a self-
reinforcing cycle in which women spend signiªcantly more time than men on housework,
time spent on housework has a direct negative effect on wages, and women engage in rela-
22                      Harvard Journal of Law & Gender                         [Vol. 28

     The “career costs” to women of interrupting work to care for chil-
dren and other family members are considerable.86 Empirical researchers
who have compared the wages of women workers with and without chil-
dren estimate a wage penalty for having children of between 10% and 15%;
this wage differential is known as the “family gap.”87 There is no similar
penalty for men; on the contrary, married men, most of whom have children,
enjoy a wage premium relative to unmarried men of 10–15%.88 Much of the
family gap among women may be explained by differences in productiv-
ity between mothers and nonmothers. Once researchers control for differ-
ences in human capital, the family gap shrinks, with the wage penalty for
having one child at about 5%, and larger penalties for the birth of addi-
tional children.89 It would be misleading, though, to rely exclusively on
estimates that control for human capital differences to discourage the re-
shaping of family leave policy—the very fact of a signiªcant productivity
gap between women who do and do not have children is sufªcient reason
for exploring policy interventions, especially given the contrasting pat-
tern for men.90
     Human capital theorists explain the effect of children on women’s
productivity in several ways.91 First, the presence of children can reduce
mothers’ human capital. Work interruptions due to childbirth and infant care
result in foregone work experience, as well as depreciation of accumu-
lated skills during the periods of interruption. If a worker switches jobs

tively more housework in part because their market wages are lower).
     86 See generally Ann Crittenden, The Price of Motherhood (2001).
     87 Jane Waldfogel, Understanding the “Family Gap” in Pay for Women with Children,

12 J. Econ. Persp. 137, 143 (1998) [hereinafter Waldfogel, Understanding the Family
Gap] (citing several studies).
     88 Id.
     89 Deborah J. Anderson et al., The Motherhood Wage Penalty Revisited: Experience,

Heterogeneity, Work Effort, and Work-Schedule Flexibility, 56 Indus. & Lab. Rel. Rev.
273, 282 (2003) (ªnding a 5% wage penalty for having one child and a 7% penalty for two
or more children); Charles L. Baum, The Effect of Work Interruptions on Women’s Wages,
16 Lab. 1, 18 tbl.4 (2002) (reporting a 4% wage penalty for birth of ªrst child and 2–3%
penalty for additional children); Shelly Lundberg & Elaina Rose, Parenthood and the
Earnings of Married Men and Women, 7 Lab. Econ. 689, 705 (2000) (ªnding a 5% wage
penalty for birth of ªrst child by married women and 9% premium for their husbands);
Jane Waldfogel, The Family Gap for Young Women in the United States and Britain: Can
Maternity Leave Make a Difference?, 16 J. Lab. Econ. 505, 516 tbl.2A (1998) [hereinafter
Waldfogel, The Family Gap in the U.S. and Britain] (ªnding a 4.57% wage penalty for one
child and a 12.60% penalty for two or more children). Each of these studies controlled for
observable factors such as education, job experience, and age, and used ªxed-effects mod-
els to control for unobserved differences, such as work commitment.
     90 See, e.g., Lundberg & Rose, supra note 89, at 706 tbl.4 (ªnding that even though

their study controlled for observable and unobserved difference, married fathers still en-
joyed a 9% wage premium for the birth of one child compared to married men with no
children, and that this premium fell to 7% for men whose wives worked relatively continu-
ously through the birth of their ªrst child).
     91 See generally Baum, supra note 89, at 3–4; Joyce P. Jacobsen, How Family Structure

Affects Labor Market Outcomes, in The Economics of Work and Family 133 (Jean
Kimmel & Emily P. Hoffman eds., 2002); Jacob Mincer & Solomon Polachek, Family
Investments in Human Capital: Earnings of Women, 82 J. Pol. Econ. S76 (1974).
2005]                 A Defense of State Paid Family Leave                              23

following a work interruption, either because her old job is unavailable or
because she needs a job with more ºexibility, some of her human capital
may be underutilized because it is speciªc to her previous employment.92
Additionally, a worker who anticipates needing to interrupt work in the
future may make fewer investments in training because she knows that
there is a limit on how much she will be able to recoup through later earn-
ings. A ªrm may also decide to invest less in training women in their child-
bearing years because it fears that it will not recoup its investment. In
sum, those who take leaves from work or quit to take more ºexible jobs
pay a signiªcant “career tax” when they do so.
     With the increased burden of household responsibilities, mothers may
also have less time to allocate to market work. They may take jobs with
greater ºexibility and lower wages on average to accommodate the new
demands on their time. Note that for men, the effects of having children
may be the exact opposite. Following the birth of a child, fathers tend to
work longer hours to support the family; this phenomenon may explain
some of the paternal wage premium. However, the capacity of fathers to
work longer hours relies on the social assumption that women should—
and the social reality that women do—take on a proportionately greater
burden of post-childbirth household labor.93
     Some economists further suggest that in addition to children reduc-
ing mothers’ productivity, women who are “less productive” may be more
likely to have children.94 Empirical researchers refer to this as a form of
“unobserved heterogeneity.” The argument is that although it cannot be
observed in conventional human capital measures, some women may be
less career oriented and more motivated by a desire to have children.95
Some may also channel signiªcant productivity into supporting their hus-
bands’ careers, e.g., planning dinners for their husbands’ work associates
and accompanying them on business trips.96
     In addition to reºecting real differences in human capital, the family
gap also reºects employers’ discriminatory assessment of workers’ pro-
ductivity. If employers cannot perfectly observe productivity, they may
estimate a worker’s productivity according to perceived work commitment,
in other words, according to stereotype. For example, employers may en-


     92 Baum ªnds that wage reductions associated with women’s work interruptions are

substantially eliminated for women who return to their previous job. See Baum, supra note
89, at 25–30. He suggests that returning to the old job allows the worker to preserve a good
job match, enjoy the beneªt of relearning old skills rather than learning new skills, retain
seniority, and continue beneªting from ªrm-speciªc human capital. Id. at 25.
     93 See Lundberg & Rose, supra note 89, at 705–06 (attributing the divergent wage ef-

fects of having children that men and women experience to increased specialization, with
mothers spending proportionally more hours in the home and fathers spending proportion-
ally more time in the workforce following the birth of a child).
     94 See, e.g., id. at 692.
     95 Id.
     96
        Jacobsen, supra note 91, at 142.
24                       Harvard Journal of Law & Gender                             [Vol. 28

gage in “statistical discrimination” against women in their childbearing
years. The fact that women in this age group are more likely than their male
counterparts to interrupt work for caregiving purposes may serve as a proxy
for an actual ªnding of a weak work commitment of a particular worker.
Employers, therefore, may be less willing to hire or train these women or
may channel them into predominately female occupations that pay less
and that have lower career prospects than those of male-dominated posi-
tions.97 Similarly, employers may make negative judgments about work-
ers who have been unemployed or employed in a part-time or temporary
job. These stereotypes may make it difªcult for a worker who has quit or
switched to a part-time job during her childbearing years to return to a
full-time, permanent job that suits her skills and preferences.98
      Gender differences in labor market outcomes also operate on a deeper
plane. Workers’ preferences may themselves be socially constructed. The
hypothesis stated above, for example, that lower average productivity may
be a preexisting characteristic of mothers, is complicated by the fact that
social norms validating women’s devotion to family over career may shape
their preferences before and after the arrival of children.99 Starting early
in life, social pressures and inºuences may skew women’s preferences to-
ward motherhood.100 Once women reach working age, their experiences or


     97 See Schultz, Life’s Work, supra note 15, at 1894–96 (arguing that women’s lower

pay is due mainly to the segregation of women into “separate-but-less-remunerative occu-
pations, ªrms, and jobs,” rather than the fact that women have more family responsibilities
and therefore select lower paying but more ºexible jobs or exert less work effort); Selmi,
supra note 1, at 721–25, 730–33 (suggesting that employers may discriminate against
women when making training opportunities available and that women occupy positions that
tend to offer fewer training opportunities, and reviewing data challenging the work effort
hypothesis and the notion that women choose more ºexible jobs).
     98 See Barry McCormick, A Theory of Signalling During Job Search, Employment

Efªciency, and “Stigmatized” Jobs, 57 Rev. Econ. Stud. 299, 308 (1990) (discussing the
“stigma” attached to occupying unskilled jobs or being unemployed between jobs and how
these situations “signal” to employers that the worker will not be committed and produc-
tive); Ian M. McDonald & Robert M. Solow, Wages and Employment in a Segmented La-
bor Market, 100 Q.J. Econ. 1115, 1124–25 (1985) (arguing that experience in secondary
employment can stigmatize a worker and may subsequently block access to primary em-
ployment).
     99 There is a huge literature on the social construction of choices. See, e.g., L. Richard

Della Fave, The Meek Shall Not Inherit the Earth: Self-Evaluation and the Legitimacy of
Stratiªcation, 45 Am. Soc. Rev. 955 (1980) (arguing that those who do poorly in the dis-
tribution of social rewards may grant legitimacy to the very social structures and hege-
monic norms that deny them status, adopting those norms as part of their self-identity);
Herbert Gintis, Education, Technology, and the Characteristics of Worker Productivity, 61
Am. Econ. Rev. 266 (1971) (criticizing neoclassical economists for assuming that educa-
tion is linked to future earnings solely due to skill development, and arguing that social
institutions such as schools affect not only the skills of future workers, but also their tastes
and personalities); George Sher, Our Preferences, Ourselves, 12 Phil. & Pub. Aff. 34
(1982) (discussing and critiquing feminist arguments that women’s traditional preferences,
which are induced by sexual stereotypes and conditioning, reinforce gender inequalities).
     100 Naomi R. Cahn, Gendered Identities: Women and Household Work, 44 Vill. L.

Rev. 525, 534 (1999) (arguing that society socializes women to want to perform household
work and childcare as well as resist working outside the home to afªrm their identity as
2005]                 A Defense of State Paid Family Leave                              25

prospective experiences in the workplace are likely to reºect and reinforce
these preferences.101 Finally, interrupting work or dropping out of the labor
market all together, for example, after childbirth, may cause depression,
helplessness, loss of self-esteem, and other psychological scars particularly
common among young women.102 Social constructivist theorists thus pos-
tulate that diminished market prospects after childbirth are a self-fulªlling
prophecy.
     It is unlikely that either the theory of human capital or that of social
construction entirely explains labor market circumstances for women.103
More plausible is that both operate together, and thus, policymakers com-
mitted to equality in the paid labor economy ought to take both seriously.
Speciªcally, in response to the human capital theory, we should promote
women’s investments in market-based human capital, encourage men to
shoulder more family caregiving obligations, and eliminate discrimination
against women workers. At the same time, we must respond to the prob-
lems presented by the social constructivist theorists by encouraging cul-
tural and attitudinal changes concerning the role of women in the home.

        C. Women’s Workforce Participation and Self-Determination

    I discussed above the idea that social norms and women’s experiences
may shape their personalities and preferences in ways that deter their par-


women); Nancy E. Dowd, Work and Family: The Gender Paradox and the Limitations of
Discrimination Analysis in Restructuring the Workplace, 24 Harv. C.R.-C.L. L. Rev. 79,
90 (1989) (arguing that work-family conºicts arise in part from socialized views of appro-
priate gender roles); Fineman, supra note 73, at 21–22 (arguing that women’s “individual
choices” are actually produced by patriarchal ideology, history, and traditions); Lucinda M.
Finley, Choice and Freedom: Elusive Issues in the Search for Gender Justice, 96 Yale L.J.
914, 935–37 (1987) (reviewing David L. Kirp et al., Gender Justice (1986), and argu-
ing that gender roles are socially constructed as opposed to a result of “natural prefer-
ences” and “free choice”).
     101 See Dowd, supra note 100, at 147 (discussing whether employers should be held re-

sponsible for sex discrimination since workplace structures play a signiªcant role in shap-
ing an individual’s perceptions of her range of employment opportunities in a gendered
manner); Vicki Schultz, Telling Stories about Women and Work: Judicial Interpretations of
Sex Segregation in the Workplace in Title VII Cases Raising the Lack of Interest Argument,
103 Harv. L. Rev. 1749, 1815–24 (1990) (arguing that both conservative and liberal theo-
ries explaining sex segregation in the paid labor economy have failed to take into account
the fact that “employers arrange work systems so as to construct work and work aspira-
tions along gendered lines”).
     102 See Knut Gerlach & Gesine Stephan, A Paper on Unhappiness and Unemployment

in Germany, 52 Econ. Letters 325, 328–29 (1996) (discussing the decreases in happiness
and satisfaction caused by unemployment in German men and women); Arthur H. Gold-
smith et al., The Psychological Impact of Unemployment and Joblessness, 25 J. Socio-
Economics 333, 334, 350 (1996) (discussing the signiªcant psychological trauma of job-
lessness, particularly among female youth); Gillian Lester, Careers and Contingency, 51
Stan. L. Rev. 73, 89–90, 116–17 [hereinafter Lester, Careers] (discussing the identity-
and ambition-shaping aspects of unemployment and employment in jobs that underutilize
the worker’s talent).
     103
         See generally Lester, Careers, supra note 102.
26                       Harvard Journal of Law & Gender                           [Vol. 28

ticipation in the workforce. Given this premise, it makes sense to probe
whether increasing women’s workforce participation could reshape and
reconstruct women’s personalities and preferences in self-afªrming ways.
      Market work can be a source of wealth, a locus of self-realization, a
source of self-esteem (either for intrinsic reasons or ºowing from the esteem
of others), a source of community membership, social contacts, and friend-
ships, an escape from isolation, and a means of structuring everyday
life.104 Clearly, the status of being a paid worker can confer certain kinds
of beneªts on the worker and can represent a good way of life. I am not,
however, defending paid work per se as a superior substantive vision of
the good life. Instead, I argue that greater involvement of women in paid
employment will enhance their capacity to determine the conditions of
their lives.

     1. Feminist Arguments Encouraging Women’s Paid Employment

     Unequal participation in paid work within a household, and thus un-
equal access to the beneªts of paid employment, creates a dynamic of
dependency—traditionally, women’s dependence upon men.105 Some femi-
nists have advocated women’s involvement in paid work on the basis of its
potential to enhance women’s personal development and self-esteem.106 In-
creasing women’s participation in full-time, paid work could give them
the independent economic means, and perhaps the conªdence, to have
bargaining power in domestic relationships, both in day-to-day decisions
that affect family life, and in the ability to walk away from oppressive rela-
tionships.107 Note that this argument based on independent economic means


      104 See, e.g., Herbert Applebaum, The Concept of Work: Ancient, Medieval,

and Modern 572 (1992) (arguing that market work is associated with status and self-
esteem and forges unique and intense human bonds); Richard J. Arneson, Is Work Special?
Justice and the Distribution of Employment, 84 Am. Pol. Sci. Rev. 1127, 1132 (1990)
(linking employment to “the satisfaction of earning one’s own keep,” self-esteem, social
bonds, and autonomy); Jon Elster, Is There (or Should There Be) a Right to Work?, in De-
mocracy and the Welfare State 53, 62–67 (Amy Gutmann ed., 1988) (emphasizing the
value of paid work and enumerating its personal beneªts).
      105 Kathryn Abrams, Choice, Dependence, and the Reinvigoration of the Traditional

Family, 73 Ind. L.J. 517, 533–34 (1998) (arguing that women’s dependency as a result of
investment in their families at the expense of investment in their own human capital un-
dermines their autonomy); see also Margaret A. Baldwin, Public Women and the Feminist
State, 20 Harv. Women’s L.J. 47, 83 (1997) (describing how the unpaid nature of domes-
tic labor reinforces women’s dependency and inferiority, leads to depression, disintegration
of self, and passivity, and diminishes their capacity to resist or organize with other women).
      106 Kimberly A. Yuracko, Perfectionism and Contemporary Feminist Values

121–28 (2003) (arguing that a married women’s paid employment advances her intellectual
and moral development and ªnancial self-sufªciency and leads to greater “self-love,” i.e.,
the belief that her own desires and ends have value independent of her husband’s); Schultz,
Life’s Work, supra note 15, at 1891–92 (describing the positive effects of increased oppor-
tunities in the paid labor force on women’s self-esteem).
      107 See Rhona Mahony, Kidding Ourselves: Breadwinning, Babies, and Bar-

gaining Power 211 (1995) (arguing that working part time makes the mother the Ҽexible
2005]                A Defense of State Paid Family Leave                              27

is not merely about the potential for paid employment to increase a woman’s
personal wealth;108 it is about the psychological independence that per-
sonal economic power confers. Taken together, these arguments posit that
women’s participation in paid employment will help them develop the conª-
dence, awareness, and economic freedom to shape their own life paths in a
fulªlling way.
     On the other hand, paid work can be monotonous, degrading, and iso-
lating. This begs the question of whether policies that increase women’s
participation in the workforce will truly increase the quality of their lives,
given that for many workers, the job available to them may be less satis-
fying than working—albeit uncompensated—in the home.109 This may be
especially true for working-class women with young children, for whom
affordable, high-quality childcare and housekeeping assistance may be
beyond reach.110 For some of these women, given the type of market work
available to them, staying home full time may be an appealing aspiration,
as devotion to family may be a source of pride and personal reward, a sign
of middle-class status, or a haven from the degradations of exploitive market
work.111 In light of these competing narratives, assuming that paid work
does in fact inºuence a woman’s personality and capacities, an important
question is whether reshaping the incentive structure so as to encourage
more women to enter the workforce for different kinds of jobs, i.e., not
just privileged jobs, is likely to lead to greater independence, self-esteem,
and moral growth.




parent,” undermining her bargaining power within the family); Yuracko, supra note 106,
at 123–26 (arguing that a full-time homemaker’s ªnancial dependency on her husband and
consequent lack of self-sufªciency so as to have no “exit option” from unsafe relationships
may breed insecurity and servility that is inconsistent with human ºourishing); Naomi Cahn,
The Power of Caretaking, 12 Yale J.L. & Feminism 177, 202–03 (2000) (describing the
ambivalence of some feminists toward the argument that women wield power in the home
through their role in caregiving, as this role ultimately puts them in a weaker bargaining
position relative to men and erodes their power); Schultz, Life’s Work, supra note 15, at
1945.
     108 The decision to marry a high-salaried man and to become a full-time homemaker may

in fact make a woman wealthier than she would be had she chosen to work. See Yuracko,
supra note 106, at 107–08 (describing and critiquing feminist arguments criticizing the
choice of homemaking on the grounds of ªnancial and emotional dependence as opposed
to economic vulnerability).
     109 See Deborah L. Rhode, Balanced Lives, 102 Colum. L. Rev. 834, 838 (2002) (stat-

ing that for many women, unpaid labor in the home is more satisfying than their paid jobs);
Joan Williams, “It’s Snowing Down South”: How to Help Mothers and Avoid Recycling
the Sameness/Difference Debate, 102 Colum. L. Rev. 812, 832 (2002) (commenting that
“[g]oing to work is no picnic for many women” and thus questioning “the point of knock-
ing oneself out, and penalizing one’s kids to boot, when work seems an uphill battle?”).
     110 Williams, supra note 15, at 157.
     111
         Id. at 156–57.
28                       Harvard Journal of Law & Gender                           [Vol. 28

     2. Psychological Studies of Work and Women’s Well-Being

      A well-known literature at the juncture of psychology and sociology
examines the relationship between work and the personality.112 A worker’s
place in an organizational structure, her opportunities for occupational self-
direction, job pressures, and the extrinsic risks and rewards built into a
job can affect the worker’s “values, self-conceptions, orientations to so-
cial reality, and cognitive functioning.”113 An important component of these
ªndings is their generalizability to other aspects of life—people who do self-
directed work, for example, come to value self-direction more highly, both
for themselves and for their children.114 The provocative implication of
this for class and social stratiªcation has garnered much attention.115
      Some researchers have examined how women’s participation in paid
work versus unpaid housework might affect their personalities. The two
types of work are characterized by different sets of structural features, which
may in turn differentially affect personality development.116 Sociologists
Chloe Bird and Catherine Ross compared the effect of various qualities
of unpaid housework and paid work on individuals’ sense of personal con-
trol.117 Sense of personal control is the belief that events in one’s life de-
pend on one’s own behavior, rather than luck or fate.118 It in turn contrib-
utes to one’s capacity for self-determination, sense of self-worth, ºexibility
in responding to stressful circumstances, ability to learn from experience
and avoid repeating past behavior that has led to bad outcomes, ability to
cope with adversity, persistence, and capacity to ward off anxiety and
depression.119 Researchers Bird and Ross found that full-time housework,
including child and eldercare, involves more actual on-the-job autonomy


     112 See generally Melvin L. Kohn, Unresolved Issues in the Relationship Between Work

and Personality, in The Nature of Work 36 (Kai Erickson & Steven Peter Vallas eds.,
1990) (summarizing research in this area, particularly the paradigm-setting work of Melvin
L. Kohn and Carmi Schooler); Kenneth I. Spenner, Social Stratiªcation, Work and Person-
ality, 14 Ann. Rev. Soc. 69 (1988) (reviewing and critiquing Kohn & Schooler’s work and
subsequent research supporting and extending it).
     113 Kohn, supra note 112, at 40–42.
     114 Id. at 42–43.
     115 See., e.g., Adina Schwartz, Meaningful Work, 92 Ethics 634 (1982) (using the

Kohn-Schooler ªndings as a basis to advocate for government intervention to change the
distribution of work so that persons across social classes have access to supervisory posi-
tions as opposed to routine jobs); Spenner, supra note 112, at 80–84 (discussing and re-
viewing evidence for thesis that parents from different socioeconomic classes exhibit dif-
ferent personality traits, values, and childrearing behavior as a result of differences in oc-
cupation).
     116 See generally Carmi Schooler, Melvin L. Kohn, Karen A. Miller & Joanne Miller,

Housework as Work, in Work and Personality: An Inquiry Into the Impact of So-
cial Stratiªcation 242 (Melvin L. Kohn & Carmi Schooler eds., 1983).
     117 Chloe E. Bird & Catherine E. Ross, Houseworkers and Paid Workers: Qualities of

the Work and Effects on Personal Control, 55 J. Marriage & Fam. 913 (1993).
     118 Sarah Rosenªeld, The Effects of Women’s Employment: Personal Control and Sex

Differences in Mental Health, 30 J. Health & Soc. Behav. 77, 79 (1989).
     119
         Id. at 79–80.
2005]                 A Defense of State Paid Family Leave                              29

than paid employment, and that greater autonomy increases one’s sense
of personal control.120 Yet, on all other measures that contribute to a sense
of personal control—degree of routinization, fulªllment, recognition from
others for the quality and value of the work, and of course pay—unpaid
housework was inferior to paid work.121 Taking all the measured qualities
into account, subjects whose primary activity was unpaid housework had
a signiªcantly lower sense of control over their lives than those whose pri-
mary activity was paid work.122
     Another important line of research suggests that the division of house-
hold labor, as opposed to simply the amount, can affect women’s psycho-
logical health. An unequal division of household labor is associated with
women’s distress and depression.123
     Most women who participate in paid employment continue to shoul-
der family caregiving obligations. This raises the question of whether com-
bining work and caregiving undermines any psychological beneªts of paid
work due to the emotional strain and exhaustion of having conºicting com-
mitments. Many studies have looked at the question of how multiple roles
affect women’s psychological and physical health. One hypothesis pre-
dicts that the strain of juggling multiple roles will lead to deleterious
health effects in worker-caregivers.124 However, there is also signiªcant sup-
port for the alternate view that occupying multiple roles can enhance rather
than deplete individuals’ energy and health, and that notwithstanding the
psychological distress of balancing work and caregiving, women who com-
bine caregiving with full-time work are happier and physically healthier than
women who do no paid work at all or who work part time.125 This may be

     120 Bird & Ross, supra note 117, at 919–20; see also Mary Clare Lennon, Women,

Work, and Well-Being: The Importance of Work Conditions, 35 J. Health & Soc. Behav.
235, 240 tbl.2 (1994) (a sample of full-time housewives reported a greater sense of auton-
omy in their work than a sample of working wives).
     121 Bird & Ross, supra note 117, at 913; see also Lennon, supra note 120, at 243

(ªnding that greater routinization of homemaking as compared with paid work offset the
psychological beneªts of autonomy).
     122 Bird & Ross, supra note 117, at 913.
     123 See Chloe E. Bird, Gender, Household Labor, and Psychological Distress: The Im-

pact of the Amount and Division of Housework, 40 J. Health & Soc. Behav. 32, 42–43
(1999) (ªnding that inequity in the division of household labor, more than the actual
amount of time spent doing housework, contributes to gender differences in depression);
Beth Anne Shelton & Daphne John, The Division of Household Labor, 22 Ann. Rev. Soc.
299, 316 (reviewing literature ªnding that both women’s time spent on housework and an
unequal division of household labor positively associated with women’s depression); Schultz,
Life’s Work, supra note 15, at 1907–09 (reviewing studies that support these propositions).
     124 Martire & Stephens, supra note 77, at 169–70 (reviewing studies showing that work

interferes with the caregiving role and vice versa, leading to psychological distress).
     125 Id. at 170 (reviewing research showing that employed caregivers tend to have better

physical health, higher self-esteem, and lower psychological distress than nonemployed
caregivers, and ªnding that part-time work does not have the same stress-reducing impact);
see also Rosalind C. Barnett & Caryl Rivers, She Works/He Works: How Two-
Income Families Are Happier, Healthier, and Better-Off 24–38 (1996) (reviewing
relevant research and contesting the notion that women who combine family caregiving with
full-time paid jobs are less healthy and happy than their stay-at-home or part-time counter-
30                       Harvard Journal of Law & Gender                             [Vol. 28

because feelings of conªdence and accomplishment in one role spill over
into other roles, that full-time employment can give family caregivers a
respite from the daily responsibilities of caregiving, and that the availabil-
ity of alternatives per se increases one’s sense of personal control by de-
creasing dependency on any one object, activity, or individual. 126 Taking this
literature as a whole, it appears that even if having multiple roles gener-
ally enhances women’s sense of personal control in many ways, excessive
role strain (e.g., working while continuing to bear a disproportionate burden
of caregiving) might undermine and offset some of these beneªts.127
     The nature of the paid work is another variable that will affect women’s
psychological health. Much of the research ªnding that paid work im-
proves women’s well-being controls for social class. When researchers have
taken social class into account, however, they have found a class gradient
of distress associated with combining work and family. Unskilled and
semi-skilled workers experience higher levels of distress associated with
balancing work and family life than skilled manual workers, who in turn
experience more distress than professional and managerial women.128 Closer
analysis of the data in these studies, then, suggests that in addition to the
existence of multiple roles, the qualitative features of the tasks or experi-
ences associated with the different roles, which are directly related to
socioeconomic class, have a signiªcant inºuence on psychological well-
being.129

parts).
     126 Martire & Stephens, supra note 77, at 170–71 (on spillover and respite); Rosenªeld,

supra note 118, at 80 (on personal control).
     127 Rosenªeld, supra note 118, at 80; Peggy A. Thoits, Multiple Identities: Examining

Gender and Marital Status Differences in Distress, 51 Am. Soc. Rev. 259, 265 (1986)
(ªnding curvilinear relationship between number of roles and psychological distress such
that individuals with very few and very many roles are more distressed than those falling in
between).
     128 Sharon Matthews & Chris Power, Socio-Economic Gradients in Psychological Dis-

tress: A Focus on Women, Social Roles and Work-Home Characteristics, 54 Soc. Sci. &
Med. 799, 803, 804 ªg.2 (2002); see also K. Jill Kiecolt, Satisfaction with Work and Fam-
ily Life: No Evidence of a Cultural Reversal, 65 J. Marriage & Fam. 23, 31 (2003) (pre-
dicting that blue-collar workers are more likely to ªnd home a “haven” rather than deriving
high satisfaction from combining work and family roles); Mary Secret & Robert G. Green,
Occupational Status Differences Among Three Groups of Married Mothers, 13 Afªlia 47,
56, 58 tbl.2 (1998) (ªnding individual well-being of married mothers in professional-
managerial occupations higher than that of blue-collar or stay-at-home mothers, and stay-
at-home mothers’ marital, parental, and family well-being higher than that of mothers with
blue-collar jobs).
     129 See Martire & Stephens, supra note 77, at 170–71 (ªnding that women with full-

time employment may enjoy greater “stress-buffering” beneªts from combining work and
caregiving than women working fewer hours due to the greater amount of time away from
caregiving); Matthews & Power, supra note 128, at 805 (showing that lack of learning
opportunity, monotony, and psychosocial strain were strongly associated with low social
class for women at age thirty-three); Secret & Green, supra note 128, at 60–62 (“The au-
thors [of cited studies] speculate that the positive relationship between multiple roles and
psychological well-being is observed primarily for mothers in professional-managerial
occupations with prestige . . . and a degree of challenge or self-direction . . . , whereas less
psychological beneªt is realized by mothers who are employed in working-class/blue-collar
2005]                 A Defense of State Paid Family Leave                                31

     Ultimately, using women’s psychological well-being to gauge the value
of policies designed to facilitate combining work and caregiving has limita-
tions. For example, these studies must compare women who have already
chosen one path or another, rather than being able to randomize subjects’
choices. As a result, selection bias may distort ªndings: it may be more
likely, for example, that people who suffer from depression stay home rather
than work because of their distress, rather than the reverse.130 On the ºip
side, the data may also understate the potential psychological beneªts of
working. Even if the balance of empirical evidence were to show that
women who combine caregiving with full-time work experience more stress
and unhappiness than those who stay home, it would not necessarily fol-
low that social policy should discourage caregivers from working. Acting
in a manner inconsistent with expected social roles can be stressful, and
therefore any effort to change social roles will involve a transitional pe-
riod of psychological conºict. In addition, looking at the relationship be-
tween role conºict and stress may unduly emphasize women’s experi-
ences during a discrete life period rather than their overall experience.
Life might be difªcult for mothers who continue full-time employment
while they have babies and toddlers at home, but in the long run these moth-
ers may experience greater psychological beneªts for having remained
active in the labor market.131
     Moreover, this analysis is complicated by a core policy circularity.
Participation in the paid labor market is a way for women to acquire inde-
pendence borne of personal economic power. But whether the conditions
prevail for work to be a positive rather than negative experience, and whether
women on the margin are likely to choose market work, will depend on
the quality of jobs available. This, in turn, may be inºuenced by work-
place beneªts such as paid family leave. Paid leave alone cannot do all
the work required to achieve this. At the same time, it has the potential, if
properly designed, to make an important and distinctive contribution.

  D. Reasons for Increasing Women’s Labor Market Attachment Redux

     The effects of caregiving obligations on women are intimately tied
to their labor market participation and to the qualitative features of their


occupations that are thought to be less emotionally rewarding and by mothers in the less-
valued homemaker role. Such ªndings are consistent with the growing consensus that ex-
perience in the work role, rather than the role itself, is the major contributor to well-being
. . . .”).
       130 Anne L. Alstott, No Exit: What Parents Owe Their Children and What

Society Owes Parents 231 n.8 (2004).
       131 Kiecolt, supra note 128, at 31. Workers with very young children are more likely to

ªnd home a haven as opposed to ªnding work a haven or ªnding combining work and
family highly satisfying. Id. However, as children get older (six to twelve years old), work-
ing mothers are more likely to ªnd combining work and family highly satisfying than to
ªnd home a haven. Id.
32                       Harvard Journal of Law & Gender                           [Vol. 28

employment in terms of pay and career advancement. Status in the paid
labor economy is an important indicator of economic and social power in
our society.132 Indeed, our laws reºect an understanding that equality in
the paid labor economy is a fundamental component of social justice. Anti-
discrimination statutes, for example, embodying core national values, are
centrally concerned with justice in the employment setting.133 The text of
the FMLA explicitly acknowledges that unequal family care burdens af-
fect the working lives of women,134 and that promoting the goal of equal
employment opportunity for women and men is consistent with the Equal
Protection Clause of the Fourteenth Amendment.135 Recently, the Supreme
Court spoke very powerfully to the importance of the preventative reme-
dial measures embodied in the FMLA for achieving gender equality in
the distribution of family caregiving obligations and paid employment
opportunities.136 If the social norm that women bear the bulk of family
caregiving obligations plays a causal role in producing inequality between
men and women in paid labor markets, then addressing this causal link is a
concern for all of society, not just for working women. Work-family poli-
cies that erode gender inequality in paid employment will in turn enhance
solidarity in a society that values such equality.
     At the same time, participation in paid employment may have sig-
niªcant psychological beneªts for women. Increasing women’s opportu-
nities to engage in market work may enhance their ability to live a life
free from dependency, and thus create the conditions for reaching their
fullest human potential.

     132 Applebaum, supra note 104, at 571 (arguing that the centrality of the work ethic in

our society means that participation in paid work is an important determinant of status and
inºuence).
     133 See, for example, Title VII of the Civil Rights Act of 1964, 42 U.S.C. §§ 2000e to

2000e-17 (2000).
     134 Family and Medical Leave Act of 1993, 29 U.S.C. § 2601(a)(5) (2000) (“[Congress

ªnds that] . . . due to the nature of the roles of men and women in our society, the primary
responsibility for family caretaking often falls on women, and such responsibility affects
the working lives of women more than it affects the working lives of men . . . .”).
     135 Id. § 2601(b)(4)–(5).
     136 Nev. Dept. of Human Res. v. Hibbs, 538 U.S. 721, 736 (2003). In this case, the

court upheld the constitutionality of Congress’s decision to abrogate state immunity from
FMLA liability on the basis of evidence of the states’ record of unconstitutional participa-
tion in, and fostering of, gender-based discrimination in the administration of leave beneªts.
Id. at 729–35. In analyzing and implicitly approving Congress’s motives for creating the
FMLA’s gender-neutral unpaid family leave entitlement, the majority wrote:

     Stereotypes about women’s domestic roles are reinforced by parallel stereotypes
     presuming a lack of domestic responsibilities for men . . . . These mutually rein-
     forcing stereotypes created a self-fulªlling cycle of discrimination that forced
     women to continue to assume the role of primary family caregiver, and fostered
     employers’ stereotypical views about women’s commitment to work and their
     value as employees. Those perceptions, in turn, Congress reasoned, lead to subtle
     discrimination that may be difªcult to detect on a case-by-case basis.

Id. at 736.
2005]                A Defense of State Paid Family Leave                              33

     Although both of these reasons might justify labor market interven-
tions designed to erode the existing differentiation in men’s and women’s
market and family roles, they may implicate different mechanisms for
ªnancing these interventions. An argument for ªnancing a social program
in a way that redistributes wealth is best supported with evidence that the
program will confer a beneªt on society as a whole rather than just on those
who receive the speciªc program beneªts. When the societal beneªts of a
program are highly targeted, it is normatively easier to justify state inter-
vention if the method for ªnancing the program involves cost-internali-
zation by program beneªciaries. In the foregoing analysis, I have offered
a basis for defending labor market intervention to increase equality be-
tween men and women in the domestic and market spheres on both grounds.
Such intervention would confer beneªts on women themselves, in the form
of their greater autonomy, and society as a whole, in the form of uphold-
ing the fundamental societal value of equality in the paid labor economy.

            IV. The Effect of Paid Family Leave on Women’s
                       Workforce Participation

     A. The Theoretical Impact of Family Leave Policies on Women’s
                         Workforce Attachment

     Although paid leave alone cannot completely equalize the division
of men’s and women’s respective contributions to family and market la-
bor, in this section I argue that both theory and data suggest that it is a
crucial piece of broader reform.
     In Part III.B, I explained how the wage gap between men and women,
as well as cultural factors, make it highly likely that if one parent decides
to leave work to stay home and care for a new child, it will be the mother.
Furthermore, once a worker has decided to quit or gets laid off after in-
terrupting work to care for a family member, it may be very costly to search
for a new job later on, especially with increased family obligations. Lim-
ited job availability and discrimination may make it difªcult for the worker
to ªnd a match between her skills and the tasks of the job so that she can
ªnd a position as good as her previous employment.137 As a result, she
may decline to reenter the workforce, or may underinvest in developing
her skills, knowing that downstream job-switching may reduce returns to
the investment.138


     137 See supra notes 92, 97, 98 and accompanying text; Jacob Alex Klerman & Arleen

Leibowitz, Job Continuity Among New Mothers, 36 Demography 145, 146 (1999) [here-
inafter Klerman & Liebowitz, Job Continuity] (discussing the labor market choices a new
mother confronts with respect to maternity leave).
     138 Employers may also decline to invest in women’s skills because they fear that they

will not recoup their investment. See supra note 92 and accompanying text.
34                     Harvard Journal of Law & Gender                        [Vol. 28

     The availability of family leave can alter these incentives, thereby in-
creasing women’s workforce attachment. Unpaid, or job-protected, leave
such as the leave that the FMLA provides, reduces the risk of having to
switch jobs after interrupting work to care for family members. Job-pro-
tected leave might, on the one hand, increase job interruptions because
women who previously worked continuously for fear of losing their jobs
will no longer face this risk. For other women, the availability of job-
protected leave will increase their willingness to invest in skills that will
advance their career with a particular employer because they know they
will be able to recoup their investment. By allowing workers to return to
their pre-leave position, job protection also reduces the degree of down-
ward mobility associated with family-related career interruptions. It might
also encourage some women to accept employment in the ªrst place if
they know that future family-related work interruptions will not jeopard-
ize their employment security.
     Adding a wage replacement component may further inºuence work-
ers’ leave decisions. Job protection is insufªcient encouragement for some
workers who need leave to actually take it.139 Compensation would likely
increase the willingness of these workers to take needed leaves. The fact
that some workers may take longer leaves than they would have taken
otherwise does not necessarily mean they will become less attached to
work or experience downward mobility in the workforce. Some women,
especially those who cannot afford not to work, will maintain a strong
attachment to the workforce with or without paid family leave. In these
cases, the availability of leave beneªts may have a positive effect regard-
less of its neutrality with respect to workforce incentives. It may reduce
the stress associated with balancing work and family, as well as improve
the welfare of children and elders who receive care from these workers.
     For other workers, the availability of paid family leave will decrease
their workforce attachment. If wage replacement makes it possible for a
worker to take a very long leave, for example, by combining paid and
unpaid leave, the worker’s workforce commitment may erode during the
extended time away. She might decide she would prefer to exit the work-
force, thus taking a longer hiatus from employment than she would have
if she were limited to unpaid leave. In addition, skill erosion as well as
the severing of her former employment relationship may mean that if she
reenters the workforce later, she will ªnd an inferior job match.
     For still other workers, we would expect wage replacement to in-
crease workforce participation. First, the availability of wage replacement in
conjunction with job-protection during family-related work interruptions
will make working (and investing in job skills) more attractive for some


     139 See supra notes 5–8 and accompanying text (discussing the phenomenon of workers

feeling ªnancial anxiety about taking leaves for family or medical reasons, and in some
cases, shortening or forgoing leave for that reason).
2005]                A Defense of State Paid Family Leave                            35

women on the margin between working and staying home or between
working and accepting public assistance. In families with children, for
example, a dual-earner arrangement, while increasing household wage in-
come, also imposes opportunity costs in terms of foregone opportunities
to give personal care and attention to those children. In addition to the
ªnancial cost of obtaining childcare, the family may also view the for-
gone opportunity of having a parent provide care personally as being
psychologically costly.140 For some workers in their childbearing years,
especially those of average or low income, the anticipation of unpaid work
interruptions in the approaching years may lead them to feel that their
economic contribution to the family through paid employment will be
outweighed by the economic and noneconomic opportunity costs of mar-
ket work. A policy of providing wage replacement during leaves of ab-
sence may tip the balance of opportunity costs in such cases. Moreover,
for reasons I address in Part V.D, the availability of the beneªt in the
form of a paid leave, as opposed to a subsidy for purchasing externally
provided childcare, such as daycare, has distinctive appeal in light of the
psychological costs of foregone parenting opportunities.
     In addition, we might predict that some women who would have quit
and switched employers if the leave were unpaid will be more likely to
return to their previous employer if the leave is paid. A worker who re-
ceives wage beneªts while on caregiving leave may feel a sense of recip-
rocity for having received paid beneªts and may be more likely to return
to the former employer as a result. Sociologists have observed that work-
ers may develop norms of reciprocity akin to those in a gift-giving rela-
tionship as a result of receiving generous treatment.141 Of course, if paid
beneªts are mandated, as this Article suggests, the worker who is atten-
tive to the sources of the various components of her compensation pack-
age will not treat the wage replacement as a gift, and this impulse to re-
ciprocate will not arise. I think it is plausible and indeed likely, however,
that for many workers, the accounting behind what makes up a paycheck
may not be so clear cut. The worker’s experience of continuing to receive
a paycheck from her employer during a period of family leave may well
have the effect of boosting her morale, commitment, and sense of loyalty
to the employer. In addition to the morale-boosting effect on workers, the
existence of “mandated generosity” may have a spillover effect on some
employers. Basic government mandates may encourage ªrms that wish to
be seen as “high road” employers to go beyond minimum compliance and
offer beneªts that exceed the ºoor. In light of this phenomenon, we might
expect paid leave, even if mandated by the government, to play a norm-


     140 Jutta M. Joesch, Paid Leave and the Timing of Women’s Employment Before and Af-

ter Birth, 59 J. Marriage & Fam. 1008, 1012 (1997).
     141 See George A. Akerlof, Labor Contracts as Partial Gift Exchange, 97 Q.J. Econ.

543, 543–44 (1982).
36                      Harvard Journal of Law & Gender                          [Vol. 28

setting role for at least some employers, and in turn trigger reciprocity
impulses that enhance employee loyalty in the aftermath of family-
related work interruptions.142
     We might also expect wage replacement to increase women’s work-
force attachment beyond job-protection mandates because some family
units will make structural adjustments in response to reduced income. If
a worker is unable to take an income-supported leave from work, the whole
family may make adjustments to accommodate reduced income, e.g., mov-
ing to a cheaper apartment or having the father take on more work re-
sponsibility.143 These adjustments may, in turn, create a new set of family
norms, expectations, and economic needs, and reduce the impetus for the
caregiving spouse to return to work following job interruption. Wage-pro-
tected family leave, because it will minimize shocks to family income
from workers’ job interruptions, may inhibit more permanent adjustments
that in the short run simply ease the strain of downward mobility, but in
the long term tend to entrench the division of men and women into their
traditional roles.
     In sum, economic theory predicts that government mandates requir-
ing paid leave for maternity purposes will affect different mothers differ-
ently: some will be unaffected, some will take longer leaves, some will take
leaves instead of quitting, and some will enter the workforce who would
not have done so otherwise. How aggregate labor supply (and demand) will
change are empirical questions. Certainly, paid leave may confer beneªts on
workers and society aside from its effects on women’s workforce partici-
pation. For example, it may have the beneªcial effects of improving par-
ent-infant bonding, and improve the health outcomes of sick children and
disabled elders. However, from the normative perspective of encouraging
women’s involvement in market work, a leave program that results in
some women taking more and longer leaves, thus decreasing their work-
force experience, will have a net positive effect only so long as it: (1) does
not actually lead them to quit or take extremely long leaves, and (2) has
the offsetting effect of encouraging other women to join the workforce or
refrain from quitting.
     The next Section, which analyzes empirical studies on the relation-
ship between family leave policies and women’s labor market behavior,


     142 See, e.g., Wen-Jui Han & Jane Waldfogel, Parental Leave: The Impact of Recent

Legislation on Parents’ Leave Taking, 40 Demography 191, 196 (2003). This study of the
impact of unpaid leave mandates in state and federal law found an increase in the likeli-
hood and duration of leaves taken by workers who were not beneªciaries of the mandate.
Id. The authors speculate that this reºects the following spillover effect: as the laws be-
came more generous, so too did ªrms covered by the laws, extending beneªts even to
workers they were not required to cover. Id.
     143 See Lundberg & Rose, supra note 89 (ªnding that fathers’ income tends to increase

following the birth of a child, possibly because fathers increase their work hours as moth-
ers spend more hours at home).
2005]                  A Defense of State Paid Family Leave                                 37

suggests that on balance, paid leave policies may enhance, rather than erode,
women’s workforce attachment.

  B. Empirical Findings on the Relationship Between Family Leave and
                   Women’s Labor Market Outcomes

     1. The Effect of Unpaid Family Leave Laws on Women’s Workforce
     Attachment

     Economists Jacob Klerman and Arleen Liebowitz compared female
labor force behavior before and after the passage of state laws in the 1980s
that provided unpaid maternity leave.144 They found these laws had posi-
tive but insigniªcant effects on new mothers’ labor market behavior.145 An
analogous study based on a more recent population sample found that the
passage of the FMLA had similarly modest effects.146 The authors of these
studies concluded that in the absence of wage replacement, parental leave
laws may do little to alter the status quo.147
     Two other studies, measuring the effects of the FMLA alone, found
that it had moderately positive effects on women’s workforce attachment.
One study that tracked mothers for a full two years after birth found that
after the passage of the FMLA, mothers returned to work more quickly
and in greater numbers, and were more likely to return to their former em-
ployer.148 Another study found that the FMLA increased employment of
women with children under the age of one.149 Taken as a whole, studies of

     144 Jacob Alex Klerman & Arleen Leibowitz, Labor Supply Effects of State Maternity

Leave Legislation, in Gender and Family Issues in the Workplace 65 (Francine D.
Blau & Ronald G. Ehrenberg eds., 1997) [hereinafter Klerman & Leibowitz, State Mater-
nity Leave].
     145 Id. at 81–82, 82 tbl.3.5 (ªnding no considerable effect of state unpaid leave laws on

new mothers’ employment levels, leave taking, or work levels during the ªrst year after
birth of a child).
     146 Han & Waldfogel, supra note 142, at 196 tbl.1, 197 tbl.2 (ªnding no signiªcant im-

pact of FMLA on incidence of leave-taking or duration of leaves taken in the ªrst three
months by mothers or fathers eligible for beneªts). These statistics controlled for state
ªxed effects, i.e., other changes in states that might have led to increased leave-taking during
that time.
     147 Id. at 198; see also Klerman & Leibowitz, State Maternity Leave, supra note 144, at

66.
     148 Sandra L. Hofferth & Sally C. Curtin, Organization for Economic Co-

Operation and Development, OECD Social, Employment and Migration Working
Paper Series No. 7, The Impact of Parental Leave on Maternal Return to Work
After Childbirth in the United States 14 tbl.1, 17 tbl.3 (2003). The study found that
after the enactment of the FMLA, on average mothers returned to work over four months
sooner than before its enactment. Id. These effects, perhaps not surprisingly, disappeared if
the sample was limited to the states (roughly half) that had a maternity leave statute in
place prior to the passage of the FMLA. Id.
     149 Jane Waldfogel, The Impact of the Family and Medical Leave Act, 18 J. Pol’y

Analysis & Mgmt. 281, 295–96 (1999) [hereinafter Waldfogel, Impact of the FMLA] (ªnding
only a 1.2% increase in women’s overall employment between 1992 and 1995, and a 7.6%
increase when considering only those women with children under the age of one in 1995).
38                      Harvard Journal of Law & Gender                           [Vol. 28

state and federal unpaid leave laws suggest that these laws led to in-
creases in both maternal leave-taking and workforce attachment, but the
effects were only modest.150

     2. The Effect of Paid Leave Laws on Women’s Workforce Attachment

     Some researchers have tried to measure the incremental effects of
adding wage replacement to family leave policies. Economist Jutta Joesch
found that women with access to paid leave were more likely to take time
off from work during the birth month than women with access to unpaid
leave only.151 However, mothers with paid leave worked longer into their
pregnancy and returned to work sooner once their infant was one month
old.152 These data suggest that in the absence of wage replacement, moth-
ers are more likely to make “all-or-nothing” choices: either return to work
immediately or signiªcantly decrease attachment to the workplace by
quitting or taking a very extended leave of one year or more.153 By induc-
ing some women who otherwise would quit or take a very lengthy leave
and possibly switch jobs to remain with their employer and return fairly
quickly, paid leave may improve their labor market attachment and status.
As we have seen, returning to work for one’s previous employer after
taking time off has the advantage of allowing the worker to exploit ªrm-
speciªc human capital, retain seniority, and relearn old skills rather than
acquire new ones.154
     A study published by the U.S. Census Bureau suggests that paid leave
increases the likelihood that a worker will return to her previous employer
rather than quitting or switching jobs.155 In a sample of women between


     150 Another study also found positive effects of leave policies on the likelihood of re-

cent mothers to return to their former employer, but is less helpful for present purposes
because it did not distinguish between paid and unpaid or public versus private coverage.
Jane Waldfogel et al., Family Leave Policies and Women’s Retention After Childbirth: Evi-
dence From United States, Britain, and Japan, 12 J. Population Econ. 523 (1999) [here-
inafter Waldfogel et al., U.S., Britain, and Japan]. This study found that after controlling
for age, education, and ªrst versus later birth, maternity leave coverage substantially in-
creases the probability that a woman will return to her prior employer following childbirth.
Id. at 536. In Britain, there was a 16% increase in likelihood that a mother would return to
her prior employer within twelve months of childbirth; in the U.S., that ªgure was 23%;
and in Japan, the chances that a mother returned to work within twenty-four months in-
crease by 73%. Id.
     151 Joesch, supra note 140, at 1016–17.
     152 Id. at 1017 tbl.4. The authors control for observable differences between groups,

such as age, education level, race, and marital status but caution that unobserved differ-
ences may explain some of the variance. Id. at 1013, 1018. For example, women who want
to return to work sooner may tend to select employers that offer paid leave policies and
vice versa. Id. at 1013.
     153 Id. at 1018–19.
     154 See Baum, supra note 89, at 25.
     155 Kristin Smith et al., U.S. Dep’t of Commerce, Maternity Leave and Em-

ployment Patterns: 1961–1995 19 (2001), available at http://www.census.gov/prod/
2001pubs/p70-79.pdf (last visited Nov. 18, 2004).
2005]                 A Defense of State Paid Family Leave                              39

1991 and 1994, among women who returned to work by the twelfth month
after giving birth, those who returned to their pre-birth employer were
more likely to have received paid leave (61%) than unpaid leave (48%) or
no leave, i.e., quit (5%).156 Conversely, among women who returned to
work within a year but switched jobs, most had previously quit (63%),
but a sizeable proportion had been on an unpaid leave (29%).157 Only 12%
of “switchers” had done so following a paid leave from their pre-birth
employer.158 The study found that workers who switched jobs had lower
hours, lower pay, and jobs demanding fewer skills than workers who re-
turned to their previous employer.159
     These studies suggest a strong association between paid leave poli-
cies and rapid returns to work. What is difªcult to discern from these studies
is the direction of causality. Are women with access to private employer-
paid leave policies on average more motivated and career oriented than
those whose jobs or employers do not provide such beneªts, or does the
provision of paid leave itself increase workforce commitment and moti-
vation?
     Studies of paid maternity leave policies in Europe are helpful in this
regard. In Europe, paid leave beneªts are government mandated, elimi-
nating the possibility of selection bias between comparison groups. Euro-
pean studies cannot compare paid and unpaid leave policies because the
norm for many years across European countries has been paid leave. How-
ever, the fact that European countries have been experimenting for sev-
eral decades makes it possible to study the long-term effects of paid leave
policies on workforce composition.
     Economist Christopher J. Ruhm studied the economic effects of man-
dated paid parental leave in nine European countries over the twenty-
four-year period of 1969 through 1993.160 During that time, government-
mandated leave entitlements roughly tripled in generosity.161 Ruhm found
these changes associated with a 3–4% increase in women’s employment


    156  Id. ªg.6.
    157
         Id.
     158 Id.
     159 Id. at 18 tbl.K. For example, 23.7% of those mothers who found a different em-

ployer experienced a pay reduction, while only 4.1% of those who returned to previous
employers had a pay reduction. Id. Similarly, while 21.2% of mothers with new employers
experienced a drop in skill level required for their work, only 1.8% of mothers returning to
their previous employer experienced a drop in skill level. Id. See also Hofferth & Cur-
tin, supra note 148, at 18, 19 tbl.4 (sampling working mothers who gave birth between
1984 and 1997, and ªnding that those who returned to the same employer within two years
after having a child earned about two dollars more per hour than those who switched jobs).
     160 Christopher J. Ruhm, The Economic Consequences of Parental Leave Mandates:

Lessons from Europe, 113 Q.J. Econ. 285 (1998) [hereinafter Ruhm, Economic Conse-
quences of Parental Leave].
     161 Id. at 295, 296 ªg.Ia (ªnding that entitlements grew from an average of ten to an

average of thirty-three weeks of job-protected leave, and from an average of seven to an
average of twenty-two weeks of fully paid leave).
40                      Harvard Journal of Law & Gender                           [Vol. 28

levels.162 He also suggests that the increase would still be substantial even
after adjusting for the likelihood that some women who would not other-
wise work temporarily enter the workforce solely to trigger beneªt eligi-
bility,163 and the likelihood that some women, while on parental leave, are
counted as “employed” though they are physically absent from work.164
Ruhm also analyzed the effects of paid leave laws on the employment levels
of women in their childbearing years as compared with older women and
found the effects to be concentrated in the younger cohort: for women aged
twenty-ªve to thirty-four years, forty weeks of job-protected paid leave
increased employment-to-population ratios by around 7–9% compared
with an approximately 4% increase for all women.165 Similarly, economists
Cal Winegarden and Paula Bracy found that as the generosity of public
paid leave programs in seventeen member countries of the Organisation
for Economic Co-operation and Development (“OECD”) increased over a
thirty-year period, so too did the workforce participation of women in
their childbearing years.166
     A recent OECD analysis also concluded that paid family leave poli-
cies increase workforce participation for women in their thirties.167 The
report grapples with the issue of causality, pointing out that in countries
where women are more present in employment, they may be better posi-
tioned to agitate for beneªts, making the causality run from increased at-
tachment to stronger policies, rather than the reverse.168 Yet the report is
skeptical that this alone explains women’s employment growth in coun-
tries with strong policies, pointing out that many countries with currently
high levels of female workforce participation, in particular, the Nordic


     162 Id. at 304 tbl.IV, 305, 311 (ªnding a 3.1% increase in women’s employment-to-

population ratio for policies permitting up to twenty weeks of paid leave, as compared with
no leave, and a 4.2% increase for policies permitting up to forty weeks of paid leave). The
analysis controlled for state-ªxed effects, i.e., other changes in states (such as the growth
of subsidized childcare) that might have increased women’s employment during that time,
although Ruhm acknowledges that it might not have captured them fully, meaning these
numbers may slightly overstate the employment increase. Id. at 311.
     163 Id. at 312–13 (suggesting that this phenomenon might account for anywhere be-

tween a 0.4% and 1.0% increase in women’s employment). Note, importantly, that some of
these “opportunistic entrants” will ultimately decide to remain in the workforce, offsetting
at least some of this discount, but Ruhm does not attempt to measure the latter phenome-
non.
     164 Id. (suggesting that this phenomenon may explain between 25% and 50% of the in-

crease in women’s employment associated with longer entitlements, but should have a
smaller effect for shorter entitlements).
     165 Id. at 310 tbl.VII, 311.
     166 See Winegarden & Bracy, supra note 69, at 1029 tbl.III (estimated marginal effect

of an added week of leave ranges from about 0.60 to 0.75 percentage points in the labor
force participation rate for women ages twenty to thirty-four, with signiªcance rates that
are uniformly very high). The authors found no support for the possibility of reverse causa-
tion. Id. at 1030.
     167 Organisation for Economic Co-Operation and Development, supra note 81,

at 153.
     168
         Id.
2005]                 A Defense of State Paid Family Leave                                41

countries, were among the ªrst to introduce work-family reconciliation
policies, prominently including paid family leave programs as part of a
deliberate effort to increase female employment levels.169
     Finally, if one’s goal is to increase workforce attachment, there may
be a limit on how generous a paid leave policy ought to be. The Joesch study
found that among women with access to paid leave, the amount of leave
available affected how much time they took: with more leave available,
women took longer leaves, although this effect eventually leveled off.170
Leaves of very lengthy duration might lead to a loss of work experience
and depreciation of human capital.171 Perhaps consistent with this, a study
of social attitudes in countries with family leave beneªts of varying gen-
erosity found that countries with extremely generous parental leave poli-
cies (Germany, with more than eighteen months leave, and Austria, with
more than one year leave, though both with only a portion at full pay)
tended to have the most “traditional” attitudes about working mothers.172
     In sum, empirical studies suggest that paid leave policies increase
the likelihood that women will take leave. At the same time, modestly gen-
erous leave policies appear to hasten women’s return to work and in-
crease the likelihood that they will return to their former employer. Such
returns are associated with higher pay, greater use of skill, and more hours
of work. Looking at labor market supply generally, paid leave policies
appear to increase women’s overall labor market participation.

    V. Objections to Government Subsidized Paid Family Leave

     There are a number of potential objections to state-subsidized paid
family leave policies. These include concerns about child and elder wel-
fare, the efªciency and justice of redistribution to those who make costly
family planning choices, equity between working and stay-at-home par-


    169  Id.
    170  Joesch, supra note 140, at 1017.
     171 Ruhm, Economic Consequences of Parental Leave, supra note 160, at 314–15; see

also Jeanne Fagnani, Parental Leave in France, in Parental Leave: Progress or Pit-
fall? 79 (Peter Moss & Fred Devin eds., 1999) (arguing that France’s policy of allowing
up to three years of leave, taken overwhelmingly by women, leads to erosion of job skills
and may reinforce employer prejudice regarding women’s work commitment); Waldfogel,
Understanding the Family Gap, supra note 87, at 142 (suggesting that Germany’s policy of
permitting up to three years of maternity leave, including a period in which women are
prohibited from returning to work, has increased the wage gap between single women and
single men and between married women and married men).
     172 See James W. Albrecht et al., A Cross-Country Comparison of Attitudes Toward

Mothers Working and Their Actual Labor Market Experience, 14 Labour 591, 597 tbl.3,
598 (2000). The authors qualify their conjecture to this effect, noting that the direction of
causality is uncertain. I will also note that the attitudes polled were with respect to either
part-time or full-time work without distinction. This leaves open the possibility that some
societies have positive attitudes toward mothers working part time but negative attitudes
about them working full time, a posture that I would argue is still fairly traditional.
42                     Harvard Journal of Law & Gender                         [Vol. 28

ents, and concerns that workplace accommodations will reinforce, rather
than erode, gender inequality. I will discuss each in turn.

                           A. Child and Elder Welfare

     Some whose principal concern is child and elder welfare may argue
that whatever beneªts paid leave policies may have once women decide
to work, encouraging women to work in the ªrst place will have a nega-
tive net effect on the lives of their children and the elders for which they
care. There is an extensive literature comparing outcomes of children whose
mothers work during the ªrst year of the child’s life with those whose moth-
ers do not. Several studies in the 1980s and early 1990s found that ma-
ternal employment during a child’s ªrst year was correlated with negative
cognitive and behavioral outcomes that may persist through the later child-
hood years.173 However, studies of the relationship between maternal em-
ployment and child outcomes will tell a misleading story unless they also
account for quality of childcare and for maternal, child, and family char-
acteristics. For instance, adverse effects appear to be concentrated in fami-
lies of moderate income.174 These same families have been found to re-
ceive the lowest-quality external childcare, raising the question of whether it
is maternal employment per se, or the quality of available childcare, that
causes the adverse effects.175 There is also some evidence that mothers who
work hardest during a child’s preschool years have lower educational at-
tainment;176 this factor may be correlated with fewer resources on the part of
these mothers to improve their children’s cognitive attainment. Finally,
maternal employment may actually improve child outcomes in the longer
term by increasing family income.177 In sum, it is far from clear that ma-
ternal employment, as opposed to independent factors correlated with ma-
ternal employment, causes negative child outcomes.
     More to the point, though, is that even if subsidies allowing women
to stay home full time improved child and elder outcomes relative to paid
leave policies, they would not advance the goal of optimizing between satis-
fying child and eldercare needs and enabling women to choose full par-
ticipation in paid work. The single-earner model will tend to reinforce
women’s decision to remain in the home or accept jobs with inferior long-
term career prospects to achieve the ºexibility of moving in and out of
the workforce.

     173 See Jane Waldfogel, Child Care, Women’s Employment and Child Outcomes, 15 J.

Popular Econ. 527, 536–39 (2002) (reviewing American and European studies on the
impact of childcare and maternal employment in the preschool years on child outcomes).
     174 See id. at 537.
     175 Id. (explaining that moderate income families tend to receive the lowest-quality

childcare, likely because they lack the income of wealthier families and the subsidies of
poorer families that often allow them to purchase high-quality childcare).
     176 See id. at 538–39.
     177
         Id. at 529, 536–37.
2005]                 A Defense of State Paid Family Leave                                43

B. General Objections to Family Subsidies: Savings Failures and Costly
                      Family Planning Choices

     Two general themes sounding against redistribution of the kind that
would be involved in a paid leave program involve savings failures and
the justice (or injustice) of redistribution to assist families with child and
eldercare needs. Neither argument is particular to the case of paid leave. The
ªrst could be leveled against any state insurance scheme, and the second
could be leveled against any subsidy directed at families.
     The savings failure objection is that paid family leave would be un-
necessary if workers saved their earnings in a responsible fashion. Perhaps,
then, to the extent that poor savers’ circumstances are the result of their own
imprudence, they should pay for their own bad planning. Arguments for
redistribution tend to be more politically palatable when the beneªciaries of
a wealth transfer are needy for reasons beyond their control.178 A related
objection is that subsidized or cross-subsidized wage replacement will dis-
courage careful ªnancial planning. Studies have shown that the existence
of public insurance can “crowd out” private insurance coverage.179
     What these objections understate is how difªcult it can be to distin-
guish between those whose need results from imprudence and those whose
inability to save is beyond their control.180 Policies designed to make poor
savers bear the consequences as a matter of desert will probably be over-
inclusive. Furthermore, we currently see very poor savings among Ameri-
cans notwithstanding the dearth of ªnancial support for family leave.181 In
other words, even if funding family leave reduces private saving, it does
not have far to fall.
     Another class of objectors focus on the justness of redistribution to
those who make costly family planning choices. For these critics it is not
obvious why the state—or citizens who do not themselves have child care-
giving needs—should pay taxes (directly or indirectly) to buffer others from
the ordinary private hardships that accompany the decision to have chil-
dren. For the most part, people choose to have children, and arguably should


     178 See, e.g., Christina Fong, Social Preferences, Self-Interest, and the Demand for Re-

distribution, 82 J. Pub. Econ. 225, 227–28 (2001) (reviewing theoretical and empirical
literature arguing that people may prefer more redistribution to the poor if they believe that
poverty is caused by circumstances beyond individual control); Seana Valentine Shiffrin,
Egalitarianism, Choice-Sensitivity and Accommodation, in Reason and Value: Themes
from the Moral Philosophy of Joseph Raz 270, 272–73 (R. Jay Wallace et al. eds.,
2004) (noting consensus among egalitarians and nonegalitarians alike that a fair distribu-
tive scheme would require individuals to internalize the costs of their voluntary choices).
     179 David M. Cutler & Jonathan Gruber, Does Public Insurance Crowd Out Private

Insurance?, 111 Q. J. Econ. 391 (1996).
     180 See supra notes 62–64 and accompanying text (suggesting that savings failures may

well be beyond the control of some and perhaps many, workers who are unable to take an
adequate family leave).
     181
         See supra note 62.
44                       Harvard Journal of Law & Gender                            [Vol. 28

then undertake the costs of rearing them.182 Furthermore, caregiving (for
children in particular) can be seen as having very powerful private
beneªts. Caring for children is usually a source of personal joy and satis-
faction, perhaps commensurate with the joy a childless person could ex-
perience given additional resources to take time away from work to travel,
nurture friendships, or pursue personal projects.183 This might make man-
dated subsidies especially troubling, or perhaps even offensive, to those
who are excluded from this pleasure, e.g., gay and lesbian couples in
states where they are not eligible to adopt children.184 Also, one’s children
may be a source of security, as adult children often look after the needs of
their parents in retirement. Finally, some are troubled by the notion of the
state providing public subsidies for children while still treating childrear-
ing choices as purely private, demanding no further accountability for the
manner in which children are raised.185 Related to these arguments is the
concern that paid family leave will simply encourage people to have more
children, thus commandeering further subsidies from nonparents.
     The arguments for requiring family caregivers to internalize the costs of
eldercare on the basis of choice stand on shakier ground. This is in part
because children do not choose to have parents and do not always will-
ingly choose to undertake their care.186 In addition, although some caregivers
will experience intrinsic beneªts from giving care to elders, for many, the
personal gratiªcation associated with eldercare is lower than for raising
children; indeed, in some cases the former is clearly more burdensome
than gratifying.187 Finally, elders themselves, in contrast to children, have

      182 See Elinor Burkett, The Baby Boon: How Family-Friendly America Cheats

the Childless 20–21 (2000) (arguing that raising children should be the responsibility of
those who bear them); see also Case, supra note 73, at 1782–84 (arguing that those who
make alternative choices shouldn’t have to subsidize those who choose to have children,
both because it will undermine prudent individuals’ decisions to refrain from having chil-
dren when they cannot afford them, and because those who choose not to be parents owe
no debt to those who do).
      183 See, e.g., Case, supra note 73, at 1767 (describing societal lack of sympathy for the

idea that a childless worker should have paid time off to pursue a valuable and sustaining
hobby rather than to have a baby); Franke, supra note 73, at 190–91 (discussing purely
personal (as opposed to altruistic) reasons mothers have children, and criticizing the socie-
tal tendency to contrast “altruistic” mothering with “selªsh” private consumption).
      184 See, e.g., Fla. Stat. ch. 63.042(3) (2003) (stating that “[n]o person eligible to

adopt under this statute may adopt if that person is a homosexual”). The Eleventh Circuit
recently upheld this statute against challenge on due process, privacy, and equal protection
grounds. See Lofton v. Sec’y of the Dept. of Children & Family Servs., 358 F.3d 804 (11th
Cir. 2004).
      185 See, e.g., Case, supra note 73, at 1772–73 (noting that parental leave mandates in-

clude no “quality control” component to monitor how people actually use their leave time);
Franke, supra note 73, at 191–92 (raising the example of parents who seek subsidies to
educate their children outside the public school system based on their rejection of the val-
ues of toleration, equality, and humanity taught in the public schools).
      186 See Jacobsen, supra note 91, at 153–54. In addition, many caregivers are daughters-

in-law who, when they chose their life partner, may not have contemplated a future of caring
for their partner’s parents.
      187
          See Martire & Stephens, supra note 77, at 168 (canvassing literature on potentially
2005]                 A Defense of State Paid Family Leave                                45

had an opportunity during their lifetimes to save and plan for future
needs, and if successful in planning, some means to purchase care during
their time of need.188 One still might argue, however, that adult children’s
obligations to their parents are long-anticipated duties ºowing from moral
reciprocity for past beneªts received, and that adult children should there-
fore shoulder the costs themselves rather than enjoy subsidies from other
citizens.
      In contrast to these views, I see the demand for care, both for children
and elders, as a societal concern rather than purely a matter of private family
obligation. Some families or individuals will make choices that result in
higher burdens of care, e.g., by having many children, and I do not make the
strong claim that they should be able to externalize the full costs of those
choices. I do, however, support a general conception that regulatory inter-
vention involving some net distributive transfers to families for depend-
ent care is an appropriate role for the state.189 Furthermore, even supposing
that providing wage replacement for family caregiving would “reward” im-
prudent savings failures or family planning decisions, I argue that to the
extent they increase women’s workforce participation and autonomy, the
beneªts of paid family leave programs outweigh their costs. Aside from
this instrumental claim (and its necessarily subjective assessment of the
tradeoffs involved), there may also be noninstrumental arguments for redis-
tribution, even to persons whose needs stem from actions within their con-
trol.190
      Empirical research suggests that maternity leave policies do increase
the fertility of workers.191 Projections of the effect of maternity leave poli-
cies on fertility in the entire population (as opposed to just working women),
however, suggest that as maternity leave policies become available, more
women are likely to enter into the workforce, and for those women, fer-



deleterious psychological and physical health effects of providing care to a disabled elder).
For a vivid personal account of how difªcult and unpleasant it can be to give care to a
disabled parent, see Case, supra note 73, at 1754 n.5.
     188 See Michael Selmi, Care, Work, and the Road to Equality: A Commentary on Fine-

man and Williams, 76 Chi.-Kent. L. Rev. 1557, 1561 n.15 (2001) (commenting that
eldercare raises distinctly different issues from childcare given that many of the elderly are
capable of taking care of themselves, either ªnancially or physically and mentally).
     189 Note that this is not purely a transfer to traditional families. Partnered but unmar-

ried workers, including gay men and lesbians, may use paid leave to care for disabled par-
ents. In addition, my proposal for paid leave would also cover care for opposite- and same-
sex domestic partners.
     190 See, e.g., Shiffrin, supra note 178, at 284–95 (arguing that accommodation, even

assuming it involves subsidizing others whose needs are partly or wholly the result of cho-
sen behaviors, is consistent with egalitarianism because it enhances all individuals’ free-
dom to make choices that express certain values, feelings, and relations).
     191 Susan L. Averitt & Leslie A. Whittington, Does Maternity Leave Induce Births?, 68

S. Econ. J. 403, 414–16 (2001) (ªnding that widespread adoption of guaranteed maternity
leave policies in the United States would increase births among young, working women,
although it would not affect the probability of having a ªrst child).
46                      Harvard Journal of Law & Gender                           [Vol. 28

tility will decline.192 This latter effect should offset the increase in fertil-
ity among women already in the workforce. Indeed, a recent study exam-
ining a large sample of countries over a twenty-year period found no signiª-
cant net effect of paid maternity leave beneªts on total fertility rates.193

           C. Equity Between Working and Stay-at-Home Parents

      Some who support the notion of subsidizing family caregiving nev-
ertheless depart from my position because they believe that fairness re-
quires any state subsidies designed to offset the costs of family caregiv-
ing to be available equally to working and stay-at-home parents.194 A core
aspect of my defense of paid leave, as mentioned, is that it would en-
courage the entrance and retention of women in full-time employment as
compared to low-hours part-time employment or full-time caregiving.195 I
do not oppose across-the-board beneªts to families with caregiving needs.
There may be sound reasons to assist families, especially those burdened
with unusually large caregiving expenses, e.g., a disabled child. But it is
central to my proposal that such subsidies be an alternative to distinct, tar-
geted subsidies for working caregivers because across-the-board caregiver
subsidies, standing alone, would not necessarily change the relative at-
tractiveness of paid versus unpaid work. Afªrmatively boosting women’s
incentives for market workforce participation is necessary to counter the
skew of existing cultural and economic incentives196 toward women com-
promising their workforce attachment to bear the disproportionate house-
hold share of unpaid family labor.197

     192 Winegarden & Bracy, supra note 69, at 1020, 1029–34.
     193
         Gauthier & Hatzius, supra note 18, at 300–01 (ªnding paid maternity leave beneªts
did not signiªcantly affect total fertility rates in a sample of twenty-two OECD countries
between 1970 and 1990, but paying families direct cash beneªts upon the birth of a child
had a positive and signiªcant effect on fertility).
     194 See, e.g., Alstott, supra note 130, at 142–45 (criticizing work-family policy advo-

cates who unduly emphasize work as the route to self-realization); Noah Zatz, Beyond
Employment: Work Requirements, Caretaking, and Liberal Justice (unpublished manuscript on
ªle with author) (arguing that limiting redistributive beneªts such as childcare subsidies to
working caretakers is unjust because it fails to recognize parental care in the home as work
on par with market work).
     195 I am also wary of proposals to make paid leave available to workers with low work-

force attachment because relaxed eligibility requirements would not provide adequate incen-
tives for meaningful workforce attachment. See supra notes 170–171 and accompanying
text and infra Part VII.F.
     196 I discuss these cultural and economic forces at length in Part III.B, infra.
     197 Professor Alstott would limit the use of subsidies to enhance caregivers’ opportuni-

ties to combine childrearing with other projects (e.g., caregiver support accounts to be used
only for childcare, education of caregiver, or retirement savings of caregiver). See Alstott,
supra note 130, at 191–95. Although Professor Alstott’s caregiving subsidies would not be
an actual cash equivalent, I believe they would offer considerably weaker incentives than
paid leave for ultimately altering the traditional division of household labor. For example,
although some women might use their subsidy to purchase childcare while working full
time, the option of using the caregiver subsidy to augment retirement savings would also
reduce the opportunity costs of foregoing paid employment to engage in full-time unpaid
2005]                  A Defense of State Paid Family Leave                                  47

      The problem with proposals that would compensate women for house-
work is similar. Such proposals take a variety of forms, but their common
thread is a reaction against the devaluation of unpaid household labor and
the failure to recognize it as a crucial component of economic productiv-
ity.198 A thorough analysis of the very complex problems raised by these
proposals is beyond the scope of this Article, but the idea that policies
aimed at encouraging women’s labor force participation will ultimately de-
value household labor must be examined.199 Social policy should not fore-
close women’s option to stay home full time, nor assume that paid mar-
ket work is intrinsically superior to unpaid domestic labor. However, a
policy such as paid leave that tips the balance of incentives need not force a
singular substantive outcome. It must be understood against the backdrop
of current incentives and opportunities that unduly discourage women care-
givers from pursuing a diverse set of contributions both within the family
and the market. The opportunity for greater diversity of experience is
essential to women’s freedom from dependency on the family as the sole
locus of identity, responsibility, and moral awareness. Remunerating pri-
mary caregiving would not transform it into market work.200 It would re-

caregiving. I am skeptical whether support accounts offering this degree of ºexibility
would signiªcantly alter women’s opportunity sets in light of existing social pressure and
economic incentives to remain in the home.
     198 See, e.g., Martha M. Ertman, Commercializing Marriage: A Proposal for Valuing

Women’s Work Through Premarital Security Agreements, 77 Tex. L. Rev. 17, 29–37 (1998)
[hereinafter Ertman, Commercializing Marriage] (arguing that premarital security agree-
ments would secure repayment of the primary wage earner’s debt to the homemaker upon
dissolution of a marriage, thus protecting women and children upon divorce and commodi-
fying women’s domestic work); Martha M. Ertman, Love and Work: A Response to Vicki
Schultz’s Life’s Work, 102 Colum. L. Rev. 848, 850–53 (2002) (arguing that focusing
only on the importance of wage labor risks devaluing women’s domestic contributions, and
that we ought to instead compensate women for housework); Katharine Silbaugh, Turning
Labor into Love: Housework and the Law, 91 Nw. U. L. Rev. 1 (1996) [hereinafter Sil-
baugh, Turning Labor into Love] (illustrating a number of ways in which the law devalues
domestic labor and advocating for a systematic attempt to value houseworkers’ contribu-
tions to family wealth on par with those of wage-earners); Katharine B. Silbaugh, Mar-
riage Contracts and the Family Economy, 93 Nw. U. L. Rev. 65 (1998) [hereinafter Sil-
baugh, Family Economy] (advocating nonenforcement of premarital agreements as a way
to counter alternate tendency to devalue non-monetary contributions to the household
economy relative to monetary contributions); cf. Nancy C. Staudt, Taxing Housework, 84
Geo. L.J. 1571 (1996) (arguing that taxing women’s housework would serve to recognize
the social and economic value of unpaid household labor).
     199 See Silbaugh, Family Economy, supra note 198, at 94–97 (distinguishing the char-

acterization of housework as creating opportunity costs for women (“loss thinking”) from
its characterization as a valuable contribution (“gain thinking”), and arguing that loss thinking
reinforces the devaluation of housework).
     200 Contra Martha M. Ertman, Changing the Meaning of Motherhood, 76 Chi.-Kent

L. Rev. 1733, 1750 (2001) (arguing that remunerating primary caregiving would make it
market work). Professor Ertman is, nevertheless, sensitive to the concern that compensat-
ing unpaid domestic labor will create incentives to reinforce traditional roles. See, e.g.,
Ertman, Commercializing Marriage, supra note 198, at 19–20, 75–76 (discussing gender-
neutral compensation of housework through security agreements, in which homemakers
receive a security interest in marital property produced by their wage-earning spouses, and
suggesting that such compensation could create an incentive for men to dissuade women
48                       Harvard Journal of Law & Gender                            [Vol. 28

main one kind of work, with the combined force of social norms and
economic incentives inhibiting more free-ranging opportunities for women
to forge a satisfying life path, whatever that path may be.201

              D. “Accommodationism” as Reinforcing Inequality

     In contrast to compensating carework within the family, public pro-
vision of caregiving resources (e.g., public daycare) would increase the
range of options available to workers who have work-family conºicts, and
for some even make it possible to work without interruption, possibly ena-
bling stronger workforce attachment than leave policies. Indeed, some
scholars have raised the objection that because they will be utilized mainly
by women, policies that accommodate worker-provided care (as opposed
to externally provided care) may merely reinforce traditional family roles.202
     As the availability of adequate external care, such as high-quality
daycare, is essential to the viability of the dual-earner family model, I
strongly support public investment in and improvement of external care-
giving institutions.203 External institutions, however, cannot substitute for
paid leave. Although the availability of quality childcare, for example, might
be necessary to encourage returning to employment after taking a leave
following the birth of a child, it is not sufªcient to prevent at least some
mothers from exiting the workforce immediately after the birth. For many
parents, daycare will not be an acceptable alternative to parental care and
bonding in the early months of an infant’s life. For acute but temporary
family health events or the birth or adoption of a child, many workers seek
the ªnancial means to take personal time away from work to tend to these
needs.204 In other words, regardless of whether parental care is superior to
externally provided care, workers may prefer the opportunity to give care
personally. Paid leave policies target this speciªc demand in a way that
external care provision cannot, and therefore may increase the workforce
attachment of a signiªcant subset of workers who might otherwise avoid,
limit, or truncate a career in the paid workforce. Supporting these choices
is good social policy, both as a pragmatic response to the many workers’


from embracing traditional roles to relieve themselves of the security interest encumbering
their income, induce men to shoulder more domestic tasks, and lead men to encourage
their female mates to engage in market labor).
     201 In addition, working exclusively in the home, even if compensated, is unlikely to

bring the enhanced personal development beneªts of market work. See supra Part III.C.
     202 See, e.g., Schultz, Life’s Work, supra note 15, at 1954–57 (criticizing accommoda-

tionists for being relatively untroubled by the segregationist implications of reforms, in-
cluding maternity and family leave, that would encourage intermittent workforce participa-
tion by women).
     203 For a thoughtful discussion and critique of public childcare in the United States, see

Gornick & Meyers, supra note 12.
     204 See Williams, supra note 15, at 48–54 (arguing that there is a strong norm of pa-

rental care in the United States, and that this partly explains the failure of full-scale com-
modiªcation of domestic work through policies such as daycare).
2005]                 A Defense of State Paid Family Leave                               49

strong desire to engage personally in some family caregiving, and as a
way of permitting, and indeed validating, the role duality of worker-
caregivers.
     One problem is that offering paid family leave as a response to the
demand for personal caregiving time assumes that workers (especially
women workers, who currently engage in far more carework than men) in
fact want this option, and that denying them this choice will discourage
their workforce participation. However, there are at least some workers for
whom the “freedom” to take personal time off is more burdensome than lib-
erating. This may be especially true in the context of eldercare.205 Cer-
tainly many workers will perceive the increased opportunity to interrupt
work to give personal care to ill family members as a beneªt that improves
their quality of life. For others, though, the very availability of the option
will make it harder to resist feelings of guilt or pressure from a spouse,
parent, or sibling to undertake a task they feel is unpleasant. Thus, for exam-
ple, wage replacement may effectively deny the adult child of a disabled
elder the relief from caregiving that a less accommodating policy would
provide.206 My proposal cannot entirely avoid this risk.207
     One solution could be to narrow the kinds of support available, for ex-
ample, by eliminating the eldercare leave option. To do so, though, would
be shortsighted in the long run. Inclusion of the eldercare leave beneªt
reduces the disproportion of heterosexual families with children as beneªci-
aries of the transfer program, and thus promotes greater fairness and in-
creases the legitimacy of the program.
     There remains the valid concern that efforts to accommodate the
demands of family within the workplace will serve unwittingly to create
a “second-class workforce,” mostly women, perceived by employers as less
committed to their jobs and thus channeled into jobs with lower pay and
more limited career potential. This is not a trivial concern. As stated in
Part III, if paid leave beneªts are excessively generous, they may encour-
age career interruptions lengthy enough to undermine women’s workforce
attachment, just as some critics of accommodation fear.208 Therefore, if
we want to increase women’s workforce participation by making it easier
for them to take leaves, while at the same time protecting against the risk



    205
        See supra notes 186–187 and accompanying text.
    206 Note, though, that the FMLA excludes care of in-laws. Thus, if a paid leave policy
were to adopt FMLA rules for beneªt-triggering events, pressure to give in-law care would
not be a problem. See The Family and Medical Leave Act of 1993, 29 C.F.R. § 825.113(b)
(2004).
    207 In Part II.B, supra, I also identiªed the possible inefªciencies at stake if a worker

pressured to take leave has a higher marginal productivity (and wage) than the external
caregiver she would hire to perform the same task.
    208 See Barbara R. Bergmann, Subsidizing Child Care by Mothers at Home, 6 Feminist

Econ. 77, 84 (2000) (suggesting that long leaves would make the distinction between paid
parental leaves and subsidies for all at-home parents less cogent).
50                         Harvard Journal of Law & Gender                         [Vol. 28

of excessive human capital erosion among those who do take leave, we
must ensure that paid leave beneªts are offered in moderation.

                               VI. Financing Paid Leave

     The cost of a paid family leave policy and who will pay for it are both
issues central to my proposal. I argue in this Part that at least some of the
costs of paid leave should be spread beyond leave-takers, in part to avoid
undermining the normative goal of work-attachment that informs this
analysis, and in part in keeping with the ethical argument made earlier
that society has a broad interest in this form of social provision. Follow-
ing a brief discussion of the ªscal costs and beneªts of paid family leave
beneªts, I will explain the distinction between payroll tax and general reve-
nue ªnancing, with particular attention to the phenomenon by which em-
ployers may shift the cost of a payroll tax to employees in the form of re-
duced wages, reduced hiring, or both. Payroll tax ªnancing, depending on
how it is designed, can spread costs beyond leave-takers, but is still prin-
cipally a tax on workers as a class, rather than a broader cross-section of
society. General revenue ªnancing spreads costs more broadly, and thus
purely ethical considerations should lead us to favor at least partial use of
general revenues. Finally, I review the quite limited empirical studies of
the relationship between paid leave programs and women’s earnings, and
speculate on the interplay between the method of ªnancing and female
workforce participation.

                  A. Fiscal Costs and Beneªts of Paid Family Leave

     Most research on the costs of family leave policies has focused on
unpaid leave of the form currently mandated by the FMLA. For this rea-
son, this discussion will ªrst explain what we know about the costs of un-
paid leave and then project what we could expect if we were to add a wage
replacement component.
     Providing workers with job-protected leave imposes several mone-
tary costs on employers. First, despite relief from wage obligations, em-
ployers may still pay some direct compensation costs. Although not all em-
ployers provide health insurance coverage, those that do have been esti-
mated to pay an average of about $250 per year to continue health insurance
provision for a worker taking leave and the worker’s family.209
     Some costs, though, are indirect, and these are more difªcult to meas-
ure. As mentioned earlier, employees who take leave will experience some
depreciation of their human capital over time, possibly requiring the em-
ployer to invest in retraining them upon return. For leaves of short dura-
tion, such as the twelve-week maximum under the FMLA, the costs of hu-

     209
           Waldfogel, Impact of the FMLA, supra note 149, at 283 & n.5 (in 1993 dollars).
2005]                 A Defense of State Paid Family Leave                              51

man capital depreciation are likely to be negligible.210 For leaves of longer
duration, however, the depreciation of human capital might become an
important cost consideration.
      Reorganizing the workplace in the leave-taker’s absence presents an-
other cost. The most common employer solutions for covering the work
responsibilities of the leave-taking employee are temporarily assigning work
to other employees (which involves administrative costs and may sap
resources in other areas of production), and hiring temporary replace-
ments (which involves both hiring and training costs).211 In the Depart-
ment of Labor’s 2000 survey, most U.S. employers covered by the FMLA
reported no noticeable impact of FMLA compliance on their establish-
ments’ productivity, proªtability, and growth.212 These reports may under-
state the actual costs of job-protected leave, however, given that many
establishments covered by the FMLA already had policies in place either
voluntarily or under state law.
      Employers enjoy several compensating monetary beneªts of unpaid
leave. Although the employer will incur reorganization costs as described
above, at least some (and possibly much or all) of these costs will be off-
set by salary savings during the leave-taker’s absence. Additionally, to the
extent job-protected leave reduces employee turnover, the employer saves
on the costs of hiring and training permanent replacements. The 1990 U.S.
Small Business Administration Employee Survey estimated that it takes
an average of two and one-half weeks to replace a terminated employee and
that the termination costs averaged at least $1,100.213 The employer will save
on these costs if a leave policy increases the likelihood of a leave-taking
employee’s return. Employers may also beneªt from improved worker mo-
rale, which would presumably translate into increased productivity.214 Ulti-
mately, taking both costs and beneªts into account, it is difªcult to de-

     210 Dorothea Alewell & Kerstin Pull, An International Comparison and Assessment of

Maternity Leave Legislation, 22 Comp. Lab. L. & Pol’y 297, 299 (2001).
     211 See Balancing the Needs, supra note 5, at tbl.6.3 (reporting that in 2000, 98% of

FMLA-covered establishments temporarily assigned work to other employees and 41%
hired outside temporary replacement workers). Other strategies include leaving the work
for leave-takers when they return, and having leave-taking employees continue some work
while on leave.
     212 Id. at tbl.6.5 (noting that 77% of establishments reported no noticeable effect of

FMLA compliance on business productivity, 88% reported no noticeable effect on business
proªtability, and 88% reported no noticeable effect on business growth).
     213 Arindrajit “Arin” Dube & Ethan Kaplan, Paid Family Leave in California: An

Analysis of Costs and Beneªts 41 (2002) (noting estimates are larger for large busi-
nesses), http://www.laborproject.org/publications/pdf/dube.pdf (last visited Nov. 18, 2004).
     214 Jane Waldfogel, Family-Friendly Policies for Families with Young Children, 5 Em-

ployee Rts. & Emp. Pol’y J. 273, 290 (2001). Note that there could also be a loss of mo-
rale on the part of employees forced to take up the slack when coworkers take leave, but
recent employee survey data suggests that this has not been a signiªcant problem in the
ªrst decade of the FMLA. See Balancing the Needs, supra note 5, at tbl.6.5 (ªnding in
2000, only 11% of covered establishments reported negative effects, 24% reported positive
effects, and 65% reported no noticeable effects of FMLA compliance on employee mo-
rale).
52                       Harvard Journal of Law & Gender                              [Vol. 28

termine whether unpaid leave actually imposes net “disruption costs” on
employers, and if it does, to what extent.
     The provision of wage replacement, of course, introduces distinct costs.
First there is the actual cost of replacing workers’ wages. Any estimate of
the ªnancial costs of wage replacement would necessarily depend on the
generosity of the program, including coverage, duration, degree of wage
replacement, beneªt ceilings, and so on. Thus, for example, the program
might cover half the workforce, providing 50% wage replacement for six
weeks, or it might cover 80% of the workforce, providing 100% wage re-
placement for six months. Clearly there would be a wide cost difference be-
tween various options, making it difªcult to estimate “typical” beneªt ex-
penditures.
     Another factor is that availability of the paid beneªt would change cur-
rent behavior. It would be inadequate simply to count the number of workers
who currently take family leaves and compute the ªnancial costs of re-
placing those workers’ wages because adding wage replacement to exist-
ing leave beneªts would likely increase the number of leaves taken as well
as the duration of leaves. In addition, we would expect some measure of
beneªt substitution. Some employers currently allow workers to use vaca-
tion days, sick leave, or personal days to replace pay during family emer-
gencies.215 For these workers, the availability of paid family leave will
free up beneªts they previously used to ªnance family leave so they can
be used for their designated purposes.216 Worker turnover also should de-
cline further compared with unpaid leave, with attendant savings for em-
ployers.
     The discussion thus far has focused on potential costs and beneªts in
the workplace, some of which would redound to the employer, regardless
of how the program were ªnanced. But we could also expect some savings
to the state. Approximately 11% of workers on unpaid leave, and, perhaps
surprisingly, 5% of those who receive at least partially paid leave, rely on
some form of public assistance.217 The availability of paid leave beneªts
would reduce these workers’ reliance on public assistance. Health beneªts to
children, the elderly, and spouses who receive care from leave-takers would
presumably also translate into ªscal savings to the state. Finally, if paid
leave programs have the predicted effect of strengthening women’s work-
force attachment, by reducing quits and encouraging new entrants, gov-
ernments will be able to collect more income tax revenues.



     215
         Cf. Dube & Kaplan, supra note 213, at 23–24.
     216 In fact, we may see less of this than one might initially think: if family leave beneªts
provide less generous wage replacement than do vacation days (70% wage replacement for
family leave compared to 100% for vacation), some workers could still opt to take the
vacation days, forgoing their right to collect the family leave beneªt to collect full pay. Id.
at 24.
     217
         Id. at 44; see also Balancing the Needs, supra note 5, at tbl.4.8.
2005]                A Defense of State Paid Family Leave                              53

     Having discussed both costs and beneªts, I will not attempt to derive
a precise net “price” of paid family leave, in part because that would de-
pend on the generosity of a program, but also because a core predicted
beneªt is difªcult or impossible to monetize. A more equal balance in
men’s and women’s respective contributions to market and family work
may well be productively efªcient, but it is hard to assign a dollar value
to the beneªt. In any event, I do not rest my argument on the claim that
greater gender equality will be productively efªcient. Instead, I will as-
sume that, at least in the short run, a program of paid family leave will
require a net ªscal expenditure. A question, then, is who ought to shoul-
der the ªscal burden.

                            B. Financing Mechanisms

     A program of social welfare beneªts, including a paid leave program,
can be designed to be more or less redistributive.218 One way to ensure
that workers have a buffer against the risk of job loss but internalize the
costs of that protection is to force them to save part of their earnings on
an ongoing basis. The state could encourage or require each employee to
invest money into a personal account which, prior to retirement, the em-
ployee could draw upon only for certain qualifying work interruptions (e.g.,
the birth of a child). When the employee retired, any remaining balance
in his account could be rolled over into a pension. A system of purely
personal accounts modulates spending across an individual’s lifetime, but
does not redistribute between individuals except to the extent that it re-
duces the relative tax burden of those who participate in the program.
     I argued earlier that women who have greater freedom to participate
in paid employment may beneªt because of their increased autonomy.219
A system of personal family leave accounts might encourage better ªnancial
planning and reduce work interruptions by women likely to have family
care obligations.220 Any autonomy beneªt of increased workforce attach-
ment could be seen as redounding mainly to women in heterosexual rela-
tionships who have care obligations. As such, a tax subsidy just large
enough to encourage private savings might be quite appealing because for
the most part it internalizes the costs of paid leave to leave-takers them-
selves. However, as I have argued, paid family leave would confer impor-
tant beneªts on society by reducing labor market inequality. In this sense,
there is also a strong ethical basis for ªnancing beneªts in a way that re-
sults in more redistribution to leave-takers.



    218 See supra Part II.C.
    219 See supra Part III.C.
    220 A 2003 legislative bill in Nebraska proposed a scheme of this nature, but the bill

was not reintroduced in the 2004 session. L.B. 37, 98th Leg., 1st Reg. Sess. (Neb. 2003).
54                       Harvard Journal of Law & Gender                            [Vol. 28

     Two common ways to ªnance social welfare beneªts to redistribute
wealth are payroll taxes and general revenue ªnancing. Relying on pay-
roll taxes allows for the redistribution of costs between workers, and us-
ing general revenues can produce an even broader spreading of costs. The
next two sections discuss each of these mechanisms in light of the social
objectives of paid family leave.

     1. Payroll Taxes

      In the United States, social insurance programs have historically
been ªnanced using payroll taxes rather than general revenues.221 All the
major social insurance programs—Social Security, Medicare, federal long-
term disability insurance, state temporary disability insurance, and un-
employment insurance—are ªnanced in whole or in large part through pay-
roll taxes.222 Moreover, the trend is toward increasing reliance on the pay-
roll tax.223
      A payroll tax may be imposed on employers, employees, or both. The
focal point of public debates about payroll taxes often centers on whom,
as between employers and employees, should bear the tax, the assumption
being that it will have implications for wealth distribution.224 Such de-
bates, however, overlook the important issue of tax incidence. The eco-
nomic burden of a payroll tax may differ from the legal burden. Both theory
and empirical evidence suggest that whatever the statutory burden may be, a
payroll tax does not signiªcantly redistribute wealth from employers to
employees.

      221 For general discussions of the historical origins of this practice, see Beth Stevens,

Blurring the Boundaries: How the Federal Government Has Inºuenced Welfare Beneªts in
the Private Sector, in The Politics of Social Policy in the United States 123 (Mar-
garet Weir et al. eds., 1988); Marie Gottschalk, The Shadow Welfare State: Labor,
Business, and the Politics of Health Care in the United States 65 (2000); Frank
R. Dobbin, The Origins of Private Social Insurance: Public Policy and Fringe Beneªts in
America, 1920–1950, 97 Am. J. Soc. 1416, 1423–1445 (1992).
      222 Of course, the accounting after revenues are generated may be quite artiªcial. Pay-

roll taxes collected for Social Security may not fund pensions directly. The point here is
that the payroll tax is increasingly used as a method of raising revenues to support a vari-
ety of government programs. Moreover, the continuing political viability of many social
welfare programs may depend on the sustained presence of payroll taxes as a signiªcant
component of federal revenue.
      223 Cf. Jonathan Gruber, Payroll Taxation, Employer Mandates, and the Labor Market,

in Employee Beneªts and Labor Markets in Canada and the United States 183,
183 (William T. Alpert & Stephen A. Woodbury eds., 2000) (noting a rise in the use of payroll
taxes as a source of public ªnance, from 12.4% of federal revenues in 1960 to 38% in
1993); Andrew Mitrusi & James Poterba, The Distribution of Payroll and Income Tax Bur-
dens, 1979–1999, 53 Nat’l Tax J. 765, 772 tbl.3 (2000) (calculating that payroll taxes
exceeded income taxes for 42% of families in 1979 and 62% of families in 1999).
      224 See Bohrer, supra note 1, at 420 (advocating ªnancing paid family leave through

employee payroll contributions because “most businesses simply cannot afford to” pay for
it); John M. Broder, Family Leave in California Now Includes Pay Beneªt, N.Y. Times, Sept.
24, 2002, at A20 (noting that proposed employer contribution was a sticking point in the
California legislative process leading to paid leave and was ultimately eliminated).
2005]                 A Defense of State Paid Family Leave                               55

      It is commonplace for economists to assert that the “statutory inci-
dence” of a tax will typically differ from the “economic incidence” of the
tax because the party on whom the tax is imposed will try to shift its bur-
den to others.225 Thus a business faced with a costly mandate may pass
the cost of that mandate to workers in the form of reduced wages. In ad-
dition, if an activity is taxed, the party taxed may do less of that activity—
e.g., if there is a tax on wages earned, the worker may work fewer hours.226
Both demand- and supply-side phenomena operate in an equilibrium that
will adjust following imposition of the tax. Trying to project (or empiri-
cally test) the outcome of this reequilibration is an important exercise if
one is interested in the distributional consequences of a particular em-
ployment mandate.

     a. Employer Mandates and the Problem of Cost-Shifting

     In recent years a good deal of scholarly attention has focused on the
distributive implications of “employer mandates.” By this I mean the le-
gal imposition on employers of either afªrmative duties or restraints in
setting the terms and conditions of employment. For example, employers
might be required to maintain minimum health and safety standards in
the workplace, accommodate the special requirements of workers with
disabilities, or refrain from discriminating against women in wages and
hiring. Assuming mandates are costly to employers, the mandate operates
like a payroll tax on the employer.
     The economics of employer mandates have been formalized in a
seminal article by economist Lawrence Summers and others who have ex-
panded on his analysis.227 For ease of illustration, consider the simple sup-
ply-demand framework in a competitive labor market from Figure 1 below,
and assume (again, solely for illustrative ease) that employers are re-
quired to bear the entire cost of the mandated beneªt. If the government
requires employers to offer a beneªt to workers that they do not already
provide, then employers will adjust their demand for labor.228 Speciªcally,
employers’ demand for labor will shift downward by an amount equal to
the cost of providing the beneªt (reºected in a shift from demand curve
D0 to D1 in Figure 1). Workers, in turn, will see the mandated beneªt as a
form of additional nonmonetary compensation and be willing to work more.
The labor supply curve will shift downward by an amount equal to the

     225 See, e.g., Richard A. Musgrave & Peggy B. Musgrave, Public Finance in

Theory and Practice 250 (4th ed. 1984).
     226 See id.
     227 See Lawrence H. Summers, Some Simple Economics of Mandated Beneªts, 79 Am.

Econ. Rev. 177, 180–81 (1989) (formally describing the predictions I summarize in this
paragraph using a partial equilibrium analysis based on price theory).
     228 For expositional ease, the effect of the tax is illustrated in Figure 1 as a uniform

shift. Under a payroll tax the shift would most likely be a percentage shift and, therefore,
not uniform at every wage. Nothing in this analysis turns on that distinction, however.
56                      Harvard Journal of Law & Gender                           [Vol. 28

valuation that workers place on the beneªt (reºected in a shift from S0 to
either of the S1 curves, explained more fully below). Taken together,
these adjustments will lead to a new equilibrium in which wages are un-
ambiguously lower, but employment levels may increase or decrease, de-
pending on how highly employees value the beneªt. In Figure 1, a low
employee valuation is reºected by the shift to S1′, while a high valuation
is reºected by the shift to S1″.
     The net value of the beneªt—whether the beneªt is efªcient—is the
difference between how much the workers value the beneªt and the cost
of the beneªt. If workers value the beneªt by more than its cost (reºected
by S1″), there will be an increase in employment levels, and total social
efªciency (i.e., employee plus employer surplus) will increase. If they
value the beneªt at less than its cost (reºected by S1′), wages will fall by
more than they value the beneªt, leading to a decrease in employment
and a net efªciency loss.229 Whatever the efªciency gain or loss, though,
this analysis does not predict a systematic redistribution from the em-
ployer to workers.

                                                     S0
Wage                                                   S 1′
                                                              S 1″

     w0
     w1′
     w1″

                                                         D0
                                                    D1

                           Q0
                         Q1′ Q1″                               Quant. Labor

                                         Figure One


     229 Relative elasticities of supply and demand may affect the magnitude, but not the ex-

istence or direction, of the three effects described. See Christine Jolls, Accommodation
Mandates, 53 Stan. L. Rev. 223, 238 (2000); Sharon Rabin-Margalioth, Cross-Employee
Redistribution Effects of Mandated Employee Beneªts, 20 Hofstra Lab. & Emp. L.J. 311,
320 (2003) [hereinafter Rabin-Margalioth, Cross-Employee Redistribution] (cautioning
that because of this phenomenon, it is misleading to assume that a large reduction in wages
accompanied by relatively unaffected employment levels necessarily reºects a highly
efªcient mandate).
2005]                 A Defense of State Paid Family Leave                               57

     The result of the equilibrium I just described is highly dependent on
whether the cost of the mandate varies with the number or kind of worker
employed. A mandate might impose a ªxed cost on the employer regard-
less of number or kind of workers (e.g., the requirement that the employer
install a ramp at the entrance for wheelchair access). The employer may
either absorb the cost or pass it along to others, but it cannot avoid or reduce
the tax by hiring fewer or different workers. A mandate might, alterna-
tively, impose costs that multiply with each additional worker hired, akin to
a surcharge (e.g., a requirement that every worker receive health insurance).
Here, the employer might choose to hire fewer workers and perhaps ask
them to work longer hours to avoid the “beneªt premium” associated with
hiring additional workers. Or, a mandate might impose costs that vary with
the number of worker-hours, or wages paid, a common example being Social
Security taxes. Here, too, the employer might simply hire fewer workers or
reduce workers’ hours in response to the tax. Finally, a mandate might
impose costs associated with hiring a certain class of workers. The em-
ployer here may have an incentive to avoid hiring workers who fall within
that class.

     b. Unpaid Leave Under the FMLA as an Accommodation Mandate

     Some have argued, for example, that the unpaid leave beneªt pro-
vided by the FMLA, while not limited to women, acts like an “accommoda-
tion mandate” because it is in fact more likely to be utilized by women of
childbearing age than by other workers.230 Although I explained earlier that
whether the employer bears any net costs in connection with unpaid job-
protected leave is contested, if we assume that it does impose net costs,
these costs may be shifted to women of childbearing age. This is because
the disparate costs of accommodation cause the demand curve for work-
ers in the accommodated group to shift downward relative to the demand
curve for other workers. On the supply side of the equation, because only
accommodated workers receive the beneªt, they will value it more highly,
on average, than other workers. Accommodated workers will be inclined
to supply more labor at a given wage than other workers (their supply curve
will shift downward relative to the supply curve of other workers).231 As-
suming wages and/or employment can adjust freely, then, unpaid leave
will reduce accommodated workers’ relative wages, even as it may in-
crease their relative employment levels. As in the discussion above, which
assumed universally applicable mandates, efªciency will depend on how
much the workers who pay for the beneªt value it relative to its cost.


     230 See Jolls, supra note 229, at 290 (analyzing the cost-shifting likely to occur under

the FMLA); Sharon Rabin-Margalioth, Anti-Discrimination, Accommodation and Univer-
sal Mandates—Aren’t They All the Same?, 24 Berkeley J. Emp. & Lab. L. 111, 152 (2003).
     231
         See Jolls, supra note 229, at 241–42.
58                       Harvard Journal of Law & Gender                            [Vol. 28

     If wages and/or employment cannot adjust freely, we may ªnd redis-
tribution to the accommodated workers. A variety of institutional barri-
ers—e.g., legal rules and workplace norms—may prevent relative wages
and employment levels from shifting in response to a costly mandate.232
Law professor and economist Christine Jolls elaborates on the importance of
antidiscrimination laws in this regard.233 In many instances, the targeted
beneªciaries of accommodation mandates are also protected by antidis-
crimination laws. If antidiscrimination laws are effective, they will pre-
vent wage or hiring discrimination between the accommodated (and pro-
tected) group (the “disadvantaged workers,” to use Jolls’s terminology) and
“nondisadvantaged workers” who are not covered by the mandate.234 As
such, the costs of the mandate will be shifted to nondisadvantaged work-
ers.235 If antidiscrimination laws, either wage restrictions or hiring restric-
tions, are not binding, however, we would expect a different outcome. If
wage restrictions are binding but hiring restrictions are not, members of
the accommodated group will bear the costs of the accommodation man-
date in the form of reduced employment.236 If employers are free to dis-
criminate in wages but not in hiring, the situation essentially reverts to what
Summers predicted, with workers who use the beneªt paying for it in the
form of lower wages.237
     Jolls argues that even where antidiscrimination laws are in place to
prevent sex discrimination in wages, occupational segregation of women
and men will have a tendency to dampen their effect.238 Where occupational
segregation exists, the cost of providing accommodations to women can be
shifted to workers in female-dominated industries in the form of reduced
wages across-the-board.239 In the absence of male employees working at the
same jobs, proof of wage discrimination is unlikely or impossible. Jolls
makes just this prediction with respect to unpaid leave under the FMLA.
She speculates, though, that because the costs of providing unpaid leave
are difªcult to monetize, an employer will tend in the short run to shift

     232 Jonathan Gruber, The Incidence of Mandated Maternity Beneªts, 84 Am. Econ.

Rev. 622, 623, 626 (1994) [hereinafter Gruber, Maternity Beneªts].
     233 See Jolls, supra note 229.
     234 Id. at 240–54.
     235 This is true even when the cost of the mandate exceeds the value of the mandate to

its beneªciaries, although certain caveats apply, e.g., the proportion of the workforce that
is nondisadvantaged must not be too small. Id. at 249–51.
     236 See id. at 255–56.
     237 See supra Part VI.B.1.a.
     238 Jolls, supra note 229, at 284. Note that women’s relative employment levels may or

may not also fall according to this model: if the value of the mandated beneªt exceeds its
cost, women’s relative employment levels will rise, and vice versa. Other factors that might
have an effect include the ratio of disadvantaged to nondisadvantaged workers in a market
and the degree to which the costs of the mandate can be monetized. Id. at 268–71.
     239 Jolls notes that accommodation mandates in integrated markets might even increase

occupational segregation, as nondisadvantaged workers who bear the costs of the mandates
exit those markets. Id. at 270–71. This would, in turn, further erode the redistributive effect
of the mandates. See id.
2005]                  A Defense of State Paid Family Leave                                 59

the costs to workers in the form of reduced employment rather than re-
duced wages, or perhaps will not shift the costs at all, until the actual
costs of giving unpaid leaves become more quantiªable.240

     c. Shifting the Costs of Wage Replacement

     A wage replacement component would add a new cost. If paid leave
were funded by requiring the employer to pay the cost of wage replacement
each time an employee takes leave, the analysis would look very similar
to the one we just completed for job-protected leave. Due to a variety of
cultural factors, wage replacement would still be used disproportionately
by female workers of childbearing age.241 Because the cost of wage re-
placement would be “attached” to individual leave-takers who fall dispro-
portionately into a particular demographic group, employers would reduce
their demand for workers in this group.
     Many workplace beneªts, though, are ªnanced through a tax on the
wages of all workers, regardless of which workers use the beneªt. If wage
replacement were funded this way, there would be no cost difference, aside
from the ªxed disruption costs of leave, between workers who take up the
beneªt and those who do not. The employer’s demand for labor in general
would fall by an amount equal to the total taxes on payroll.
     On the supply side, workers would be willing to work more in response
to the beneªt, shifting their supply curve downward. But workers may
value the beneªt differently. Those who intend to have children, for ex-
ample, presumably value wage-replacement beneªts for family leave more
highly on average than do other workers.242 The former group of workers
would be willing to supply more labor at the going wage in response to

     240 Id. at 269–70, 291. Another phenomenon that may shift the costs of accommodation

mandates disproportionately to women is the more general tendency of mandated beneªts,
accommodation or otherwise, to entrench labor market segmentation. Employers may shift
costs disproportionately to secondary labor market workers (those who are easily replace-
able because their human capital is basically irrelevant) in an effort to avoid cutting com-
pensation of incumbent primary employees (those whose human capital is necessary to
their job). See generally Daniel S. Hamermesh, Labor Demand 46–55 (1993); Olivia S.
Mitchell, The Effects of Mandating Beneªts Packages, in Research in Labor Economics
297, 304–05 (Laurie J. Bassi et al. eds., 1990); Rabin-Margalioth, Cross-Employee Redis-
tribution, supra note 229, at 328–43. Women tend to be disproportionately employed in the
secondary sector. See Shultz, Life’s Work, supra note 15, at 1894–95.
     241 See discussion supra Part III.B.
     242 Valuation issues are actually more complex than uptake (i.e., who is likely to use

the beneªt). In addition to women of childbearing age, men whose employment provides
family leave coverage to their spouses or partners will place a signiªcant value on the so-
cial provision of family leave beneªts even if they are not themselves covered or do not
intend to take leave. For these men, the ability of their partners to take a subsidized leave is
effectively a subsidy to the whole family. Note also that although workers who do not in-
tend to have children may place no value on the ability to take parental leave, they may
highly value the ability to take leave to care for an ill spouse or disabled elder. Thus we
would predict a more even valuation of family leave beneªts across demographic groups
than this example suggests.
60                      Harvard Journal of Law & Gender                          [Vol. 28

the addition of the beneªt. Other workers would value the beneªt less
(assume no value for simplicity), and therefore would supply less labor at
the going wage, as some portion of the wage/beneªts package would then
be worthless to them. So long as the zero-valuing workers did not exit the
labor force entirely, the net effect would likely be redistribution from lower-
to higher-valuing workers. There would still be no wealth transfer from em-
ployers to workers—only redistribution between groups of workers.
      In theory, this analysis would not change if the tax were nominally
imposed on employees rather than employers. The tax would not affect em-
ployers’ demand for labor because, a fortiori, there would be no cost dif-
ference between workers. On the supply side, workers across-the-board
would now directly pay the tax in proportion to their earnings. Higher-
valuing workers would still value the beneªt by the same amount, and would
be willing to supply the same additional amount of labor in response to
introduction of the beneªt as in the earlier scenario. Others would still
value it at zero, and perhaps some on the margin would drop out of the labor
market. In sum, the same redistribution between groups of workers would
be expected regardless of whether the tax were statutorily imposed on em-
ployers or workers.243
      In practice, though, if employers nominally bore the tax, it might be
difªcult for them to shift costs to workers immediately or fully. Abrupt
wage adjustments might undermine employee morale, and thus employ-
ers might phase in wage reductions as they hire future employees. Also,
employers might ªnd it difªcult to shift the full costs of the tax to employees
if it would undermine their ability to attract workers who did not stand to
gain as much from the leave policy. Thus despite theoretical predictions,
employers and workers might partly share, at least temporarily but per-
haps also in the long run, the cost of a mandate ªnanced by a tax nomi-
nally imposed on employers.
      In addition, there may be an interaction between the availability of
wage replacement beneªts and the “disruption costs” of job-protected leave.
This interaction may in turn affect the cost-shifting dynamic. We would
predict, for example, that for some workers the availability of wage re-
placement will increase the likelihood or duration of leave. Thus even if
the state tries to reduce discrimination against “high uptake” demographic
groups in a paid leave program by using a ªnancing mechanism that com-

     243 There may be some cost differential between workers due to the fact that payroll

taxes tend to have a ceiling. Once the tax ceiling is reached, the tax behaves like a ªxed-
cost beneªt: the relative tax paid per worker begins to diverge for workers whose income
exceeds the ceiling. High-wage workers are, in essence, a better value, and as I explained
earlier, the employer may have an incentive to work them more hours rather than hire new
workers, thus redistributing from low-wage to high-wage workers. See sources cited supra
note 240. This is part of a more general problem with the regressivity of payroll taxes.
Payroll taxes are regressive because of the tendency toward ceilings on tax bases, and be-
cause at the high end, wages tend to fall as a proportion of total income. Musgrave &
Musgrave, supra note 225, at 258.
2005]                A Defense of State Paid Family Leave                             61

pletely decouples the cost of wage replacement beneªts from workforce
composition (e.g., by using general revenues), the very availability of paid
leave may increase beneªts uptake and thus costs for the employer. Si-
multaneously, though, wage replacement will prevent some workers from
quitting, thereby reducing the costs of turnover, and thus the disruption
costs of leave-taking. In sum, whatever the incremental costs of wage
replacement, its presence might either increase or decrease the disruption
costs associated with having a leave policy in place, and by extension,
the costs shifted to female workers of childbearing age.
     This Article’s defense of paid leave hinges on the empirical assump-
tion that it will encourage women to enter or remain in the workforce de-
spite competing caregiving obligations. Financing the program in a way
that leads employers to avoid hiring women in their childbearing years
risks thwarting this central goal. Because women’s labor market decisions
often turn on an assessment of the comparative opportunity costs of paid
work versus staying home, a ªnancing method that shifts costs to women
in the form of reduced wages may also undermine the program’s goal. As
the forgoing discussion explained, however, a payroll tax will not neces-
sarily create this problem. It might spread costs sufªciently between
workers to accomplish the instrumental goal of increasing women’s work-
force participation, even if accompanied by some reduction in women’s
wages. Nevertheless, I have defended paid leave as advancing the broader
goal of gender equality in paid employment. This value affects society
broadly. On this basis, there may be a distinct argument, aside from its “in-
centivizing” effects on female labor supply, for spreading the costs of paid
leave beyond workers to a broader base of society.

     2. General Revenues

     General tax revenues, either state or federal, are a common source of
funding for social insurance programs outside of the United States. For
example, many European Union countries have paid family leave programs
ªnanced at least partly by general revenues.244
     Financing through general funds distributes the economic costs of a
beneªt most broadly. Included in general revenues are all income taxes (not
just wage income), corporate taxes, excise and sales taxes, estate and gift
taxes, and payroll taxes. In essence, all taxpayers contribute to general reve-
nues in proportion to their tax burdens. Therefore, this method of ªnancing
draws revenues from a broader swath of the population than a payroll tax,
which targets employers and workers alone.



     244 See Kamerman & Kahn, supra note 17, at 211–12. (listing several European coun-

tries in which family beneªts are ªnanced at least partly through general revenues); Ruhm,
Economic Consequences of Parental Leave, supra note 160, at 291–92.
62                       Harvard Journal of Law & Gender                             [Vol. 28

     At the same time, using general funds leads to opportunity costs:
people’s ability to use their money for private uses is restricted by the state’s
decision to make a public expenditure ªnanced by general tax revenues.
Behavioral distortions, i.e., changes in the behavior of taxpayers who alter
their choices in response to the taxation of a particular activity, are a type
of inefªciency. For example, if the income tax (the largest source of gen-
eral revenues)245 is increased to help ªnance a new beneªt, there may be
some workers currently on the margin between work and leisure who inefª-
ciently exit (or enter) the workforce. But general revenue ªnancing does
not lead to distortions in the activity supported by any particular pro-
gram. It does not give employers an incentive to avoid hiring or to reduce
the wages of workers who would receive beneªts under the program.246
As such, it avoids the potential problem, identiªed by those who have stud-
ied the distributive effects of mandated employment beneªts, of hurting
those it is designed to help.
     Let me brieºy highlight the contrast between the cost-shifting analy-
sis I have just reviewed for an employer-paid mandated beneªt and what
we would expect with a scheme that is ªnanced using general tax revenues.
Suppose that employers were reimbursed by the government out of gen-
eral revenues for the full costs of providing a mandated beneªt. An indi-
vidual employer’s demand for labor in such case would be unaffected by
the mandate (the demand curve in Figure 1 would remain ªxed at D0).
Therefore, although the supply curve would shift depending on how much
workers valued the beneªt, there would be no countervailing pressure on
demand as a result of the cost to the employer of providing it.247 Wages
would not fall to the same extent as they would if employers had to pro-
vide the beneªt using ªrm revenues.248 General revenue ªnancing would
have the advantage of maximizing the base of taxpayers who share in the
costs of bringing about a more equitable distribution of men’s and women’s
participation in market and family labor.




     245 Approximately 58% of general revenues come from income taxes. Ofªce of Mgmt. &

Budget, Budget of the U.S. Government, Fiscal Year 2003, at 410 tbl.S-11 (2003),
available at http://www.whitehouse.gov/omb/budget/fy2003/pdf/budget.pdf (last visited Nov.
18, 2004).
     246 The income tax does have some distortionary effects. Any increase in the income

tax will affect the work-leisure tradeoff for some individuals. But this is not the same as a
tax that dampens the speciªc activity one is seeking to encourage.
     247 Note also that under general revenue ªnancing it becomes more difªcult to deter-

mine whether the mandate is efªcient. With an employer funded mandate, all that must be
assessed is whether the equilibrium quantity of labor rises or falls. Here, it will rise regard-
less of whether the mandate is efªcient. It will increase further, however, if the mandate is
efªciency enhancing.
     248 Relative elasticities of supply and demand would also affect wages, although this

phenomenon would not depend on the source of revenue ªnancing. Thus, if supply increased
but demand did not, wages would fall under any scheme.
2005]                 A Defense of State Paid Family Leave                               63

     3. Empirical Evidence of Cost-Shifting Effects of Paid Leave
     Policies on Women’s Wages

     I have discussed two factors related to paid leave policies that may
exert opposite effects on women’s wages. First, if paid leave policies re-
duce the number of workers who quit and increase the likelihood of leave-
takers returning to their previous employer, we would expect wages to in-
crease relative to a no-leave or unpaid leave policy. At the same time, if
the costs of leave-taking are shifted from employers to workers, we might
expect to see them manifested in the form of reduced female wages.
     Some researchers have studied whether having access to maternity
leave beneªts affects mothers’ wages. Professor Jane Waldfogel, analyzing
the labor market outcomes of two cohorts of young American women ob-
served in 1980 and 1991, found that women covered by paid or unpaid ma-
ternity leave policies (the study did not try to distinguish between them) who
returned to their pre-birth employer had higher pay, all else equal, than other
mothers who worked prior to giving birth but did not have such coverage.249
Later studies found similar results for women in Britain and Japan (both
of which have paid maternity leave statutes).250 Waldfogel and her coau-
thors controlled for observable differences between the groups with and
without maternity leave (women with coverage had higher levels of edu-
cation and pay and worked for larger and unionized ªrms), and although
there may be unobserved differences between the groups, the authors
claimed that the wage difference reºected the increased likelihood that
women with coverage would return to work for their previous employer af-
ter childbirth.251
     Other researchers have tried to measure the overall wage effects of
maternity leave legislation. Professor Jonathan Gruber studied the effects
of state laws prohibiting health insurers that provided hospital and medi-
cal beneªts from excluding maternity-related expenses.252 For employers,
this added between $750 and $1,000 (in 1990 dollars) to the cost of pro-
viding health insurance to married employees in their childbearing years.253
Gruber found that married women of childbearing age paid for most of
the costs of these mandates in the form of lowered wages.254 Note that the

     249 Jane Waldfogel, Working Mothers Then and Now: A Cross-Cohort Analysis of the

Effects of Maternity Leave on Women’s Pay, in Gender and Family Issues in the Work-
place 92, 115–23, 116 tbl.4.6 (Francine D. Blau & Ronald G. Ehrenberg eds., 1997).
     250 See generally Waldfogel, The Family Gap in the U.S. and Britain, supra note 89;

Waldfogel et al., U.S., Britain and Japan, supra note 150, at 531, 541 (ªnding that the
availability of family leave increases the likelihood that a mother will return to her pre-
birth employer and noting that this job retention may promote higher wages).
     251 Waldfogel, Understanding the Family Gap, supra note 87, at 150–51.
     252 Gruber, Maternity Beneªts, supra note 232.
     253 Id. at 626 tbl.1.
     254 Id. at 630–32, 632 tbl.3 (ªnding a 5.4% decrease in the relative wages of twenty- to

forty-year-old married women in states in which such laws were implemented as compared
with changes in this cohort’s wages in states that had not passed such laws).
64                     Harvard Journal of Law & Gender                        [Vol. 28

tax on employers in this case varied directly with beneªt usage rather
than being a general tax on payroll. As such, it acted as a surcharge on hir-
ing workers likely to use beneªts.
     Studies of the wage effects of the unpaid leave mandate under the
FMLA are more equivocal. Professors Sandra Hofferth and Sally Curtin
found that for American mothers who worked during the ªnal three months
of their pregnancy and returned to the workforce within two years post-
birth, living in a state that had a maternity leave policy at the time of child-
birth had a signiªcant negative effect on earnings.255 The same study found
that passage of the FMLA also negatively affected the change in pre-birth
to post-birth earnings.256 Another study that analyzed the effects of the
FMLA on the relative wages of full-time working mothers and full-time
working women without children in their childbearing years found no signi-
ªcant wage effects.257 Both FMLA studies looked only at wage effects in
the ªrst two years after passage of the FMLA, thus leaving unresolved
the question of the longer-run wage effects of the FMLA’s mandate. Hof-
ferth and Curtin were surprised by the negative wage effects they observed
and predicted a longer-run reversal given their other ªndings (discussed
earlier) that the FMLA hastened mothers’ return to work and increased their
likelihood of returning to the same employer.258
     Ruhm’s twenty-four-year survey of paid parental leave mandates in
Europe, discussed earlier, also examined wage effects.259 Recall that in
the period analyzed, entitlements tripled in generosity.260 The overall av-
erage wage replacement rate by the end of the period analyzed was about
80%.261 Paid, job-protected leaves of relatively short duration (ten weeks)
had little consequence for women’s wages (not just mothers), although
leaves of longer duration (twenty to forty weeks) were associated with a
2%–3% wage reduction.262 Ruhm’s study suggests that paid leave poli-
cies, at least those that permit shorter-duration leaves (somewhere between
ten and twenty weeks), can increase women’s workforce participation
without signiªcantly reducing their wages.263 The absence of a wage ef-
fect suggests that either the costs to employers of short leaves are trivial,
or that for shorter duration leaves, the positive wage effects of increased


     255 Hofferth & Curtin, supra note 148, at 18, 19 tbl.4 (multivariate analysis in

which the dependent variable was the gain in hourly earnings between birth and two years
post-birth).
     256 Id.
     257 Waldfogel, Impact of the FMLA, supra note 149, at 296–99, 197 tbl.6 (comparison

groups were men from 19 to 45 years old and women from 46 to 60 years old).
     258 Hofferth & Curtin, supra note 148, at 20.
     259 See Ruhm, Economic Consequences of Parental Leave, supra note 160.
     260 Id. at 295.
     261 Id. at 307. In most programs, wage replacement starts out at or around 100% and

levels off to a lower percentage in later weeks.
     262 Id. at 303–07, 304 tbl.4, 306 tbl.5.
     263
         See supra notes 160–165 and accompanying text.
2005]                A Defense of State Paid Family Leave                              65

workforce attachment—and the employee’s attachment to a particular em-
ployer—may offset any costs shifted to female workers.
      An important feature of European policies is that general revenue
ªnancing covers most of the cost of wage replacement beneªts. Ruhm
suggests that in the countries he studied, the main costs employers seek
to shift are the non-wage “disruption costs” of leave.264 Thus, in theory,
parents would enjoy the beneªts of paid leave policies but risk employer
cost-shifting for only a relatively minor portion of the full costs of the
beneªt.265
      A shortcoming of available studies is the lack of any systematic ef-
fort to analyze the inºuence of different paid leave ªnancing mechanisms
on wages and workforce participation. The studies reviewed above focus
on a range of ªnancing mechanisms for legislated programs, including
paid beneªt mandates on employers, unpaid leave mandates, and gov-
ernment programs in which the cost of wage replacement is substantially
paid for by the state. I found no study that tests theoretical predictions
about differential labor market effects depending on method of ªnancing.
Also unexamined in the empirical literature is the prediction that paid
leaves of very lengthy duration might lead to a loss of work experience and
depreciation of human capital akin to what we see with workers who quit
work and subsequently reenter, with the effect of depressing wages.266 It
would be useful if researchers could try to assess whether there is a “tip-
ping point” beyond which further increasing the duration of available leave
ceases to enhance attachment and instead undermines it, and whether wage
depression follows from that.
      In summary, paid leave policies may result in wage costs to workers
most likely to take leave—women in their childbearing years. This is not
necessarily bad for women. If women want paid leave beneªts, but the
market does not provide them, they may be more willing to work, even at
a reduced wage, in a regime that mandates these beneªts. In that case, the
beneªciaries of the program would essentially internalize its costs.267
Still, there is a strong normative claim that the costs of paid leave beneªts
should be more widely spread. Future empirical research should tackle
more directly and precisely two important predictions of the theoretical
literature: that leaves of very long duration, as well as ªnancing that re-
quires users of beneªts to internalize their costs, may eliminate or reduce


    264
         Ruhm, Economic Consequences of Parental Leave, supra note 160, at 289–90, 314.
    265  Ruhm also suggests that while maternity policies increase the supply of female la-
bor, demand for female labor might be relatively inelastic, resulting in a decrease in
women’s wages. Id. at 314; see also Mitchell, supra note 240, at 312–13 (speculating that
downward wage pressure might result from the increase in labor supplied by women in
response to the (then merely proposed) FMLA).
     266 See supra notes 170–172 and accompanying text.
     267 Although if the cost were shifted in the form of a reduction of all women’s wages,

women without family caregiving needs would be subsidizing leave-takers.
66                     Harvard Journal of Law & Gender                       [Vol. 28

the potential for paid family leave policies to increase women’s work-
force attachment.

                   VII. Implementing Paid Family Leave

      Given the Bush administration’s repeal of the BAA-UC, legislation
providing for paid family leave is most likely to happen at the state level.
It is plausible that programs could take a variety of forms, some incorpo-
rated into existing state beneªt programs, some created anew. For this rea-
son, rather than craft a detailed proposal for a particular paid family leave
program, it seems most helpful to suggest several core features that a paid
leave program, in whatever form, ought to have. In this ªnal part of the
Article, I will describe some examples of models for implementing a pro-
gram of paid leave. Then I will discuss ªve basic features I believe are
most important.

                         A. Models for Implementation

     A number of models have been proposed for implementing a paid fam-
ily leave program. These include expanding existing state UI programs,
expanding existing state TDI programs and creating new ones, creating
separate family leave insurance programs, providing incentives for em-
ployers to expand sick leave policies, and reimbursing employers who
voluntarily create paid leave programs.268
     It is possible to create a paid family leave program that is independ-
ent of existing social insurance programs. The beneªt of doing so is that
it would not require modifying an existing program’s beneªt triggering
events and eligibility requirements. A new program might also be able to
depart more freely from undesirable but institutionally ingrained features
of existing programs, some of which I will discuss below.269 At the same
time, expanding existing programs is more practical for a variety of rea-
sons. Building on to an existing administrative infrastructure is likely to
be more efªcient than starting a program from the ground up. In addition,
it may be more politically tractable to introduce legislation that expands
an established program. An expansion appears to make only an incremental
change along a spectrum of beneªts that have already met with public ap-
proval rather than create a new social welfare beneªt. Perhaps for these
reasons, most of the attention has been focused on the ªrst two alterna-



     268 See Nat’l P’ship for Women and Families, Family Leave Beneªts: A Menu

of Policy Models for State and Local Leaders (2001) (describing various alterna-
tives), http://paidleave.org/docs/759_PolicyModels.pdf (last visited Nov. 18, 2004).
     269 An example is the low minimum taxable wage base in UI. See infra notes 272–274

and accompanying text.
2005]                 A Defense of State Paid Family Leave                              67

tives listed: expanding UI programs and expanding or creating TDI pro-
grams to encompass paid family leave.270

     1. Unemployment Insurance

     Although BAA-UC was recently repealed, UI has been thought of as
a vehicle for funding paid leaves for many years, and may come back on
the federal legislative agenda in the future. Given that every state currently
has an administrative apparatus for assessing and disbursing UI wage
replacement beneªts, this existing framework provides a promising infra-
structure on which to graft a family leave program.
     UI provides partial, temporary wage replacement for workers who
become involuntarily unemployed. It is jointly administered by the fed-
eral and state governments. Each state has created its own UI program
within the guidelines of federal law.271 Federal law requires states to tax
employers on the ªrst $7,000 of each worker’s earnings per year and de-
posit the revenues into a trust fund reserved exclusively for paying un-
employment beneªts.272 Although many states have elected to set their tax-
able wage base at a higher level, many also remained at or near the fed-
eral minimum in 2002, including the populous states of California, Flor-
ida, and New York.273 A base of $7,000 is very modest compared with other
major social insurance programs such as Social Security, which has a
taxable wage base more than twelve times that amount.274 The upshot is
that most state unemployment insurance taxes tend to be quite regressive,
with lower-income employees paying a signiªcantly higher proportion of
their wages in UI taxes than high-income employees. Ideally, the taxable
wage base in a paid leave program ªnanced through payroll taxes would



     270 The Right Start Act of 2001 proposed to fund state efforts to develop programs for

family leave paid through UI, TDI, or other mechanisms. H.R. 265, 107th Cong. §§ 501–
533 (2001); S. 18, 107th Cong. §§ 501–533 (2001). Although the bill died in both the
House and Senate, it reºected a common view about the most promising potential institu-
tional vehicles for funding parental leave. Recent bills in twenty-one states have proposed
to expand UI. See supra note 9 and accompanying text. Recent bills in Hawaii, Maine, New
Jersey, and Washington would either expand TDI or establish new, similar insurance
schemes for family leaves. See bills cited supra note 10.
     271 The Social Security Act of 1935, 42 U.S.C. §§ 501–504, 1101–1108 (1994 & Supp.

V 1999), authorizes the allocation of funds to states that comply with federal requirements
and establishes the Employment Security Administration Account.
     272 Federal Unemployment Tax Act, 26 U.S.C. §§ 3306(b)(1)–(f) (2000).
     273 California and Florida remain at $7,000, while New York has increased the wage base

to $8,500. Although only twelve states remain at the minimum, most states still hover just
slightly above the minimum: in 2002, the median taxable wage base was $9,000 (the aver-
age was roughly $12,400). See Employment Training and Admin., U.S. Dep’t of Labor,
2002 State Taxable Wage Base and Tax Rates, http://www.workforcesecurity.doleta.
gov/unemploy/statetax.asp (last visited Nov. 18, 2004).
     274 The maximum taxable wage bases for Social Security in 2005 will be $90,000.

Cost-of-Living Increase and Other Determinations for 2005, 69 Fed. Reg. 62,497 (2004).
68                      Harvard Journal of Law & Gender                           [Vol. 28

be much higher than the $7,000 minimum required by UI—preferably
closer to the $87,900 taxable wage base of Social Security.
     On the beneªt side of the equation, employees receive wage re-
placement as a proportion (usually about half) of their pre-layoff earnings,
with ºoors and ceilings on beneªts that tend to limit beneªts to within
the average earnings range. The average weekly beneªt in 1998, for ex-
ample, was nearly $200.275 The maximum duration of beneªts is typically
twenty-six weeks, but whether an employee is eligible to collect beneªts
for the maximum duration depends, in almost all jurisdictions, on that em-
ployee’s work history. In 1998, the average duration was 13.8 weeks.276
     Federal law also provides incentives for states to use experience rat-
                                                          277
ing in their programs, and all ªfty states have done so. Experience rat-
ing varies the tax on each employer according to the costs it imposes on
                     278
the insurance pool. This encourages employers to self-regulate: if they
internalize some of the costs, they will be more reluctant to lay off work-
ers. The linkage of tax liability with beneªt use, however, could create a
signiªcant cost-shifting problem of the sort discussed in Part III. If an em-
ployee’s characteristics lead an employer to predict that she will present
a higher risk of becoming unemployed and collecting beneªts, the employer
will have an incentive to avoid hiring that worker or reduce the worker’s
wages as a compensatory measure. This might have a very salient effect on
women’s employment and wage levels. For this reason, BAA-UC sensibly
permitted states to devise programs that would exempt employers from
experience rating in the case of beneªts for birth and adoption leaves. Any
future reprise of a regulation like BAA-UC must also incorporate this
exception to prevent targeted cost-shifting.
     UI is partially designed to stabilize the economy in the face of sec-
toral shifts and possible mass dislocations of workers. It is also designed
to optimize the use of human capital, with compensation intended as a tool
to help unemployed workers engage in a more sustained search than they

     275 The average replacement rate in 2003 was 43% of wages. See Employment &

Training Admin., U.S. Dep’t of Labor, Unemployment Insurance Chartbook, at
chart A.13 (2003), available at http://www.ows.doleta.gov/unemploy/content/chartbook/
descript.asp (last visited Nov. 18, 2004); Employment & Training Admin., U.S. Dep’t
of Labor, Handbook No. 394, Unemployment Insurance Financial Data Handbook
[hereinafter Handbook] (reporting 1998 statistics), available at http://ows.doleta.gov/
unemploy/content/hdbk394_99/lnk98.html (last visited Nov. 18, 2004).
     276 See Handbook, supra note 275.
     277 See 26 U.S.C. § 3302(b) (allowing an employer a credit against its federal tax in an

amount equal to the amount by which its taxes were discounted pursuant to a system of
experience rating). See generally Wayne Vroman, Experience Rating in Unemployment
Insurance: Some Current Issues 9–20 (Unemployment Ins. Occasional Paper No. 89-6,
1989) (explaining how experience rating works).
     278 Experience rating is imperfect, meaning that the heaviest users do not fully com-

pensate the pool for the disproportionate costs they impose on the pool; in general, how-
ever, there is a correspondence between level of usage and level of tax. Elsewhere, I dis-
cuss both methods for experience rating and their imperfections. See Lester, Wealth Redis-
tribution, supra note 56, at 344 & n.32, 345 & n.33.
2005]                A Defense of State Paid Family Leave                            69

could if forced to rely on their own savings. Eligibility rules reºect the phi-
losophy that UI is a program of wage replacement insurance for stably
employed workers whose job loss is anomalous rather than characteristic,
whose unemployment is involuntary, and who are actively seeking reem-
ployment.
     Thus, although most workers are covered by UI, the major limitation
on the allocation of beneªts is whether eligibility rules are met.279 First,
UI is targeted at workers with a history of relatively stable workforce at-
tachment. To be eligible for beneªts, a worker must meet minimum hours or
earnings requirements during the previous year.280 Such requirements av-
erage in the range of about $1,760 (about eight-and-a-half weeks of work
at the minimum wage).281 The worker must also have spread that work out
over the course of the year to some minimal degree, rather than having
worked in just one quarter of the year. The effect of these rules is to screen
out workers with very low or irregular hours or earnings, especially low-
hours part-time, part-year, or temporary workers.282 Second, UI is targeted at
the worker who is involuntarily laid off and then immediately begins
searching for a new job. A claimant will ordinarily be disqualiªed if the
reason for unemployment is that she quit her job without “good cause” or
was dismissed by the employer for misconduct.283 She must also be “able
and available” for reemployment. To continue to collect beneªts, a recipient
must register with an unemployment ofªce and provide ongoing veriªcation
of an active job search.284
     Both the reason for employment interruption and the short-term goals
of the unemployed worker are very different when family caregiving is
involved. Although employers may rehire workers after a layoff trigger-
ing UI beneªts, eligibility formally hinges on job separation—layoff or
quit—rather than on leave explicitly understood as temporary. In addi-
tion, the decision to conceive or adopt a child is usually voluntary, and a
worker who has separated from work to care for a child is not “able and
available” for work. Thus compensating a worker for work interruptions

     279 As of 1995, more than 90% of civilian employment was covered. Advisory Council

on Unemployment Comp., Unemployment Insurance in the U.S.: Beneªts, Financing
and Coverage 163 (1995). However, certain groups of workers are excluded. Id. For
example, coverage of agricultural and household workers is limited, and self-employed
workers are excluded. Id. In addition, certain state and local government employers and
some nonproªt employers are exempt. Id.
     280 Econ. Policy Inst., Unemployment Insurance: Facts at a Glance (2004), at

http://www.epinet.org/content.cfm/issueguides_unemployment_facts (last visited Nov. 18,
2004).
     281 These are my calculations based on U.S. Department of Labor data. See Employ-

ment & Training Admin., U.S. Dep’t of Labor, Comparison of State Unemployment
Laws 3-5 to 3-6 (2003), available at http://www.ows.doleta.gov/unemploy/uilawcompar/
2003/monetary_entit.pdf (last visited Nov. 18, 2004).
     282 See Lester, Wealth Redistribution, supra note 56, at 347–55, for a more detailed

discussion.
     283 Id. at 350.
     284
         Id. at 352–53.
70                      Harvard Journal of Law & Gender                          [Vol. 28

caused by childcare demands would require relaxing two of the three tra-
ditional eligibility criteria—the “able and available” requirement and the
involuntariness requirement.285 For reasons discussed above, we would also
need to suspend experience rating for separations due to family leave. A
minority of states already treats workers who leave employment by rea-
son of pregnancy or work-family conºict as having “good cause” for quit-
ting and thus not disqualiªed from collecting UI beneªts.286 Despite the
fact that some of the changes necessary to cover family leave under a UI
scheme are feasible, the conceptual ªt between UI and family leave wage
replacement program designed to encourage leave-takers to return to their
former employer remains quite imperfect.287

     2. Temporary Disability Insurance

     TDI provides partial wage replacement for workers who ªnd them-
selves unable to work due to a temporary nonwork-related illness or in-
jury. Five states—California, Hawaii, New Jersey, New York, and Rhode
Island—plus Puerto Rico currently have TDI programs. Each program in-
cludes pregnancy- and childbirth-related temporary absences as insured
events, but with the recent exception of California, does not currently
include absences for infant care beyond the immediate period of physical
recovery from birth. Each state funds its program through a payroll tax,
in some cases on both workers and employers, and in others on workers
alone.288 Four of the ªve states (all but Rhode Island) have a “pay-or-
play” method that allows employers to satisfy the statutory requirements
through self-insurance. In these instances, however, employers are pro-
hibited from requiring employees to make contributions that exceed what
they would be required to contribute under the state plan, and beneªts
must not fall short of what would be available under the state plan.
     The taxable wage base in TDI programs tends to be higher than for
UI: the median taxable wage base across the ªve states with TDI pro-
grams in 2002 was $35,495 (compared with $9,000 for UI).289 Beneªts


     285 BAA-UC permitted modiªcation of the “able and available” requirement. See 20 C.F.R

pt. 604 (2000). The involuntariness requirement has always been a creature of state law.
     286 See sources cited supra note 34.
     287 See Lisa Bornstein, Inclusions and Exclusions in Work-Family Policy: The Public

Values and Moral Code Embedded in the Family and Medical Leave Act, 10 Colum. J.
Gender & L. 77, 114 (2000) (noting that the inconsistency between the involuntary unem-
ployment premise of UI and the fact that most maternity leave is voluntary poses a phi-
losophical problem for proposals to use UI to fund family leave); Drew, supra note 1, at
383–84 (raising a similar criticism); Ulrich, supra note 1, at 39–40 (echoing this concern).
     288 All ªve existing state programs allow mandatory contributions from employees. All

but California and Rhode Island also require contributions from employers. See Ofªce of
Research, Evaluation and Statistics, Soc. Sec. Admin., Social Security Programs
in the U.S. 48 (1997); Steven K. Wisensale, Family Leave Policy 120–21 tbl.5.1
(Daniel J.B. Mitchell ed., 2001).
     289
         Author’s calculations are based on examination of current statutes in TDI states
2005]                 A Defense of State Paid Family Leave                               71

vary across states, but as with UI, beneªt amounts and duration vary ac-
cording to an employee’s prior earnings history. Also like UI, the replace-
ment rate is about half of an employee’s weekly wages, with speciªed
ºoors and ceilings similar to those for UI.290 The average weekly beneªts
for the ªve states with TDI programs range from $143 to $351.291 The
maximum duration of beneªts ranges from twenty-six to ªfty-two weeks,292
but in 1997, the average duration of leave actually taken was about eight
weeks.293 The taxable wage base for TDI programs is signiªcantly higher
than for UI, while the wage replacement rates are roughly comparable,
suggesting that TDI programs are more progressive than UI.
     As with UI, coverage is very broad, making the eligibility determina-
tion the major screening device. Here, too, workers must satisfy a minimum
past earnings or hours requirement to qualify for beneªts.294 In contrast to
UI, continued eligibility hinges on the continuation of the disabling con-
dition, rather than ability and availability for work.
     Disability insurance exists to provide ªnancial relief to a worker
who is physically incapacitated from work for a temporary period due to
illness or injury. Like UI, the program generally contemplates a worker who
did not choose to interrupt employment—a condition that the worker who
decides to conceive or adopt a child would not clearly satisfy—although
existing programs have been required to cover pregnancy-related incapacita-
tion since the late 1970s.295 Still, parental leave extending beyond the
period of pregnancy-related incapacitation, as well as family illness leave,
would both involve decisions that turn on needs external to the worker’s
own physical capacity, such as a decision to care personally for an ailing
parent or a newborn infant rather than contract with a professional care-



(California $46,327, Hawaii $35,495, New Jersey $23,500, New York $6,000, Rhode Is-
land $45,300). The average taxable wage base for TDI programs in 2002 was $31,324.
     290 The ºoors range from $14 (Hawaii) to $68 (New Jersey), and the ceilings range

from $170 (New York) to $504 (Rhode Island). Temporary Disability, supra note 31, at
8-5 to 8-6.
     291 Soc. Sec. Admin., Social Security Bulletin, Annual Statistical Supple-

ment 312 tbl.9.C (2000) (relying on 1997 statistics), available at http://www.ssa.gov/
policy/docs/statcomps/supplement/2000/9c.pdf (last visited Nov. 18, 2004).
     292 Temporary Disability, supra note 31, at 8-5 to 8-6.
     293 Soc. Sec. Admin., supra note 291. Data for New Jersey were unavailable and thus

not included in this calculation.
     294 For details, see Temporary Disability, supra note 31, at 8-4 to 8-5.
     295 The Pregnancy Discrimination Act of 1978 (“PDA”) requires that workers affected

by “pregnancy, childbirth, or related medical conditions” be treated “the same for all em-
ployment-related purposes, including receipt of beneªts under fringe beneªt programs, as
other persons not so affected but similar in their ability or inability to work.” Pub. L. No.
95-555, 92 Stat. 2076 (codiªed as amended at 42 U.S.C. § 2000e(k) (2000)) (amending
Title VII, which applies only to employers with ªfteen or more employees). The passage of
the PDA was in response to General Electric Co. v. Gilbert, 429 U.S. 125 (1976), in which
the Court rejected a claim that an employer’s disability insurance plan discriminated on the
basis of sex when it excluded pregnancy and related conditions from otherwise compre-
hensive coverage.
72                    Harvard Journal of Law & Gender                    [Vol. 28

giver.296 So, as with unemployment compensation, there are some incon-
sistencies between the typical disability leave and the coverage contem-
plated under paid family leave. The ªt between parental leave and a dis-
ability leave program nevertheless seems more comfortable than the ªt
with unemployment compensation because both parental leave and TDI
programs aim to accommodate employees who are temporarily not seek-
ing reemployment and will likely return to their former employers.
     Note that whether grafted onto a UI or TDI program, introduction of
family leave beneªts would create a covered event for which there is signiª-
cantly higher moral hazard than for the risks currently covered. Given a
subsidy, some individuals will take leaves who would not otherwise have
done so, and some will take longer leaves. Indeed, this is one of the in-
tended goals of adding a wage replacement component to family leave.
There is no reason to believe that one program is more vulnerable to
these types of moral hazard problems than the other.
     In terms of both conceptual ªt and practical design features, then,
the TDI model seems more adaptable than the UI model to the aims of a
paid family leave program. In general, TDI programs tend to have a higher
taxable wage base than UI programs (meaning there is greater progres-
sivity of the tax and beneªt schedule), and would require less dramatic
modiªcations of the basic features of the program than if UI were adapted to
accommodate family leaves (e.g., there is no experience rating and no
“able and available” requirement in TDI programs). The “pay or play” con-
cept in several of the TDI schemes is particularly attractive in its capacity
to give employers the opportunity to experiment with private plans, pos-
sibly offering greater incentives for some employers to set an example by
offering programs that exceed the minimum. Only ªve states currently
have TDI programs, but several administer them through the state UI in-
frastructure. Thus it would be possible to create a paid family leave in-
surance program that tapped state UI administrative infrastructure with-
out being subject to the constraints of the UI program itself.
     As an example, consider the new California Family Temporary Dis-
ability Insurance (“FTDI”), which adds family illness and nonmaternity
parental beneªts to the existing State Disability Insurance (“SDI”) beneªts
of personal illness and maternity leave.297 The SDI program, although
conceptually distinct from UI, is operated through the state’s UI adminis-
trative apparatus. The new beneªts are ªnanced with a fairly low tax on
employees’ wages. In 2004 and 2005, it will be .08% of their wages up to
the taxable wage limit of $68,829 in 2004 and $79,418 in 2005; it is sub-
ject to further increase in later years, up to a maximum of 1.5%.298 This is

     296
        See supra Part II.B.
     297Cal. Unemp. Ins. Code §§ 3300–3306 (Deering 2004).
    298 Id. § 984(a)(2)(B); Ca. Employment Dev. Dep’t, Paid Family Leave Insurance:

Frequently Asked Questions [hereinafter Frequently Asked Questions], at http://
www.edd.ca.gov/direp/pºfaq2.asp (last visited Nov. 18, 2004).
2005]                 A Defense of State Paid Family Leave                                73

an average of about $70 per employee per year.299 The program covers
about 70% of California’s workforce,300 with eligibility requiring wage
earnings of at least $300 in a twelve-month base period prior to applica-
tion for beneªts.301 It provides wage replacement at approximately 55%
for up to six weeks,302 with weekly beneªts in 2004 ranging from a mini-
mum of $50 to a maximum of $728.303 Personal illness and maternity
leave beneªts in California’s preexisting state disability insurance pro-
gram provide up to ªfty-two weeks’ wage replacement (on the same terms
as above).304

                                      B. Financing

     I have argued that there are both practical and ethical reasons to
spread the costs of paid family leave beyond leave-takers themselves.
Both across-the-board payroll taxes (i.e., taxes based on some percentage
of all workers’ wages rather than beneªt usage levels) and general reve-
nue ªnancing are able to do this. As explained, payroll taxes spread costs
across a smaller base, workers, than general revenue ªnancing. However,
most major social insurance programs in the United States, including all
of the TDI programs, are ªnanced through payroll taxes, perhaps reºecting
greater political acceptability of this route as compared to general reve-
nue ªnancing. Furthermore, during periods of ªscal retrenchment, pro-
grams funded through general revenues tend to be cut before programs
funded through payroll taxes.
     If ªnanced through a payroll tax, there is a question of whether the
nominal tax should fall on employers, workers, or both. I mentioned that
although in theory a nominal tax on employers will be passed fully to work-
ers, in practice, employers may ªnd it difªcult to pass the tax fully to
workers. This leads to fuller cost-sharing, at least in the short run, than if
the tax is imposed entirely on workers. At the same time, imposing the
nominal tax at least partly on workers may have stronger political appeal
and facilitate buy-in by the public and pro-business legislators.305 As I have


     299 Workers currently pay a tax of 0.9% of their income, or up to $417 per year, to

cover state disability insurance taxes. See Frequently Asked Questions, supra note 298.
     300 This ªgure was derived from calculations using data from Ca. Employment Dev.

Dep’t, Quick Statistics, at http://www.edd.ca.gov/eddquickstats.asp (last visited Nov.
18, 2004), and Ca. Employment Dev. Dep’t, Fact Sheet: Disability Insurance Pro-
gram 1 (2003) [hereinafter Fact Sheet], at http://www.edd.ca.gov/direp/de8714c.pdf (last
visited Nov. 18, 2004).
     301 Temporary Disability, supra note 31, at 8-5.
     302 Frequently Asked Questions, supra note 298.
     303 Cal. Unemp. Ins. Code §§ 2652–2653, 2655, 3301 (Deering 2004); Fact Sheet,

supra note 300 (2003).
     304 §§ 2626, 2653, 2655.
     305 See Broder, supra note 224 (noting that shared payroll taxes were a sticking point in

the California FTDI, and limiting the tax to workers was one of the compromises that led
to its passage).
74                         Harvard Journal of Law & Gender                          [Vol. 28

argued, cost internalization by the class of individuals covered by the
beneªt may be acceptable, and perhaps appropriate, as long as the degree
of cost-shifting to leave-takers is not so great as to undermine the work
incentive effects that I argue are so central to the value of paid leave. An
across-the-board payroll tax, by tolling workers while at the same time
spreading costs beyond the subclass of workers most likely to take leaves,
may strike a reasonable balance in this regard. There should also be a
relatively high ceiling on the taxable wage base, closer to what we see
for Social Security than for UI, to avoid regressivity.306
     I have also argued that at least some degree of general revenue
ªnancing is ethically justiªed. States may be open to funding paid leave
at least partly through general funds: in the wake of BAA-UC, several
states considered bills for paid family leave that would be ªnanced in part
through direct general fund allocations307 or state tax credits.308 Thus gen-
eral revenue ªnancing may be more politically plausible than some com-
mentators have suggested.309 One appealing suggestion is to use a general
fund allocation to reimburse employers based on how many workers ac-
tually take leaves, and in doing so, reduce the likelihood of employers
shifting “disruption costs” to leave-takers in the form of targeted wage or
employment reductions.310
     A program of paid family leave insurance, then, could be funded by
a payroll tax on employers, workers, or both. The resulting trust fund could
also be augmented with general tax revenues, although it seems unrealis-
tic and perhaps imprudent to advocate for a system funded entirely through
general funds.

                                  C. Wage Replacement

     Recall that ªnancial constraints prevent some workers from taking
leaves, and workers who do take leaves tend to have higher incomes and
salaries than those who do not.311 Consistent with this, we might expect

     306   See supra notes 271–274 and accompanying text.
     307
           See Leave Initiatives, supra note 9 (Illinois, Massachusetts, Minnesota, and Ver-
mont).
     308 See id. (Colorado, Hawaii, Massachusetts, and Missouri).
     309
         See, e.g., Issacharoff & Rosenblum, supra note 1, at 2215 (concluding, in regard to
the use of general revenues to pay for family leave, that “there is little purpose in suggest-
ing beneªts approaches whose political viability approaches absolute zero”); Glavinovich,
supra note 1, at 172–73 (citing Dan Cordtz, Hire Me, Hire my Family, Fin. World, Sept.
18, 1990, at 79 (stating a similarly skeptical viewpoint)); Grill, supra note 1, at 391 (criti-
cizing general revenue ªnancing because it would lead employees to feel entitled to beneªts
rather than having earned them); Jolls, supra note 229, at 280–81 (noting that subsidizing
employers for the cost of providing the accommodation may be stigmatizing and lack po-
litical durability).
     310 Issacharoff & Rosenblum, supra note 1, at 2218–21 (suggesting this as part of a

proposal to expand UI to paid maternity leave); see also Ulrich, supra note 1, at 52–53
(offering a similar proposal).
     311
         See supra note 7 and accompanying text.
2005]                A Defense of State Paid Family Leave                              75

the relative opportunity costs of paid work versus staying home to be a
closer draw for middle- and lower-middle-income caregivers.312 Although
lower-income workers may be more likely to feel that they have to work,
paid leave may affect the relative opportunity costs of paid employment
versus welfare.313 Furthermore, salaried, higher-income, and more-educated
workers are currently more likely to receive wage replacement during family
leaves than nonsalaried, lower-income, and less-educated workers.314 The
gap appears to be widening.315 For these reasons, we might want a paid
leave program designed more assuredly to reach workers of middle and
lower incomes.
     There are a number of ways to structure beneªts aimed at progressiv-
ity while also satisfying competing concerns. I recommend beneªts paid as
a percentage of pre-leave earnings with a ºoor and ceiling imposed, rather
than ºat rate payments based on average wages, as some have proposed.316
Because they are less likely to under- or overcompensate workers whose
wages deviate from the average, wage-proportional beneªts are better suited
than ºat beneªts to sustaining workers’ standard of living and avoiding
the necessity of major structural changes to the family economy during
leave.317 For this to hold true, the percentage of wage replacement should
be fairly generous, perhaps 70% of pre-leave earnings. Less than full
wage replacement is also classic check against moral hazard; a slight but
not excessive drop in earnings allows workers to avoid major shocks to
their standard of living while also minimizing frivolous leave-taking and
maintaining work incentives. One risk of providing only partial wage re-
placement is that lower-earning workers will receive beneªts that are too


     312 For a discussion of the comparative opportunity costs of working versus full-time

caregiving, see Part IV.A, supra.
     313 Recall that almost 9% of leave-takers in 2000 ªnanced such leaves through public

assistance. See supra note 8.
     314 Balancing the Needs, supra note 5, at tbl.A2-4.1 (according to a 2000 survey,

87.6% of salaried workers who chose to take leave for family or medical reasons received
wage replacement compared to 54% of those who were paid hourly; 80.9% of leave-takers
with a college education received some wage replacement compared to 39.8% of those
with less than a high school education; 81.2% of leave-takers with incomes between $75,000
and $100,000 received some wage replacement compared to 26.2% of those with incomes
less than $20,000).
     315 Brooks Pierce, Compensation Inequality, 116 Q. J. Econ. 1493 (2001) (noting that

employer expenditures on paid leave declined for low-wage labor and increased for high-
wage labor between 1981 and 1997, and that this widening nonwage compensation gap
exceeded the wage compensation gap during the same period).
     316 The Washington state family disability insurance bill set ºat rate beneªts at $250

per week. H.B. 2399 § 7(1), 2004 Leg., 58th Sess. (Wash. 2004.); S.B. 6272 § 7(1), 2004
Leg., 58th Sess. (Wash. 2004); see also Mahony, supra note 107, at 250 n.3 (describing
paid leave beneªts that depend on pay as regressive and unfair).
     317 See Kamerman & Kahn, supra note 17, at 210 (describing earnings-related beneªts

as more closely linked to an expectation of continued wage work); Lundberg & Rose, su-
pra note 89; supra text accompanying note 143 (discussing the gender equity importance
of reducing the need for structural change to the family economy as an adjustment to non-
supported leave).
76                     Harvard Journal of Law & Gender                         [Vol. 28

low to enable them to take leave, and that the beneªt may therefore be
skewed toward middle- and upper-income workers. A ºoor and ceiling on
beneªts ensures provision that is both sufªcient at the low end and simul-
taneously limited at the high end where workers are most likely to have
an adequate personal safety net or private employer beneªts. In addition,
exempting lower- income workers from the payroll tax would help avoid
the potential for regressivity in the program.318

                                    D. Eligibility

     The goal of UI is to encourage more stable employment. Consistent
with this, UI eligibility requires substantial workforce attachment, typi-
cally measured by a worker’s hours or earnings during a base period prior to
unemployment. These eligibility requirements provide incentives for sta-
ble employment and prevent workers with relatively low workforce at-
tachment from using UI to subsidize intermittent workforce participation.
TDI, by contrast, may be available to workers who have relatively mini-
mal work hours in the preceding period. This makes sense in light of
TDI’s mission of providing income security in the event of disability-related
work interruption.
     The eligibility requirements that are most appropriate for a paid fam-
ily leave program turn on how the goals of the program—increasing
women’s workforce attachment and the likelihood that a worker who takes
leave will return to her former employer—can best be achieved. Increased
workforce participation might take a variety of forms: it might mean
more women with any type of job, be it full-time or part-time, or with
single or multiple employers. The sociological literature reviewed in Part
III contains some evidence that full-time work has a stronger correlation
than part-time work with well-being in women who combine work and
caregiving.319 The value of job continuity with one employer, as opposed
to just full-time work, is more central in the economics literature, which
ªnds a link between continuous attachment to one employer and greater
career advancement, wages, and skill utilization.320
     Taken together, the evidence suggests that although the best scenario
for maximizing advancement and wages is full-time, stable attachment to
a single employer, higher work hours per se are preferable to low or no work
hours. Consequently, strict work requirements, in terms of either work
hours, job stability, or both, may be more appropriate than relaxed work re-
quirements. However, strict work requirements may be in tension with the
progressivity goals discussed in the previous section on wage replace-


    318 See, e.g., Ulrich, supra note 1, at 52 (recommending this kind of exemption for a

program of paid family leave insurance).
    319 See supra Part III.C.2.
    320
        See supra notes 154, 159 and accompanying text.
2005]                 A Defense of State Paid Family Leave                                77

ment. The poorest workers are also the most likely to occupy temporary
and part-time jobs, move between employers, and have low work hours.
Reconciling this tension is a central challenge in thinking about how to
design eligibility and wage replacement rules for a paid leave program.
      California’s FTDI program, as we have seen, has very relaxed eligi-
bility rules. Workers with earnings as low as $300 during the twelve-month
base period are eligible for beneªts.321 At the state minimum wage, this is
equivalent to forty-four hours, or about one week, of work in the space of
a year. Such relaxed eligibility rules may increase women’s incentives to
shift to or remain in part-time rather than full-time employment. This in
turn might undermine, rather than advance, the career-building and gen-
der equity goals that paid leave programs have the potential to facilitate.
      At the other end of the spectrum, the FMLA restricts eligibility to
workers who have at least one year of service with a single employer of ªfty
or more employees, and whose work hours for that employer averaged
twenty-four hours or more per week (which is about 60% full time) in the
year preceding leave.322 This strict requirement of stable attachment to
one employer protects employers against the costs of guaranteeing job
security for very recently hired leave-takers. At the same time, it excludes
many other stable workers who would beneªt from job-protected leave, such
as the signiªcant number who work for ªrms with fewer than ªfty em-
ployees which, on the whole, have lower wages and less generous family
leave beneªts than larger establishments.323 Moreover, the requirement of
stable attachment to one employer, while perhaps reasonable in the context
of mandating job guarantees, may be less important for purposes of man-
dating wage replacement ªnanced by an across-the-board payroll tax.
      Recognizing the special advantage of job continuity with one em-
ployer, but also recognizing the value of increasing work hours per se,
eligibility requirements for paid family leave insurance should fall some-
where between the two extremes of FTDI and FMLA just presented. While
we should be wary of extending beneªts to part-time workers with very

    321  Temporary Disability, supra note 31, at 8-5.
    322
         Recall that the eligibility requirements for the unpaid leaves available under the
FMLA are one year of service, 1250 hours worked in the year preceding leave, and an
employer with ªfty or more employees working within seventy-ªve miles of the jobsite.
     323 Eliminating the small employer exemption has allowed FTDI to cover almost one

half of those workers who are not covered under the FMLA. Natalie Koss, The California
Family Temporary Disability Insurance Program, 11 Am. U. J. Gender Soc. Pol’y & L.
1079, 1084 (2003). The wages of employees steadily climb as establishment size increases,
with the difference between the extremes—average weekly wages in private establishments
of fewer than ªve employees ($644) as compared to establishments of one thousand or
more employees ($1,154)—being nearly two-fold. See Bureau of Labor Statistics, U.S.
dep’t of Labor, Quarterly Census of Employment and Wages (author’s calculations
using 2004 data in this interface), at http://data.bls.gov/labjava/outside.jsp?survey=en (last
visited Nov. 18, 2004). A 2004 survey of California establishments found that establish-
ments with 200 or more employees were 2.1 times more likely than smaller establishments
to provide family and medical leave beneªts beyond those required by law. See Milkman
& Appelbaum, supra note 49, at 6.
78                       Harvard Journal of Law & Gender                           [Vol. 28

low hours of work, this does not mean that only workers with full-time
hours should be eligible. Part-time work may be the gateway to fuller work-
force participation, and inclusion of higher hours part-time workers may
be a good way to facilitate this “foot-in-the-door” potential. In addition,
eligibility of higher-hours part-time workers might encourage men, over
time, to reduce their own work hours to play a more active role in family
caregiving. What minimum requirement will optimize the balance between
reducing incentives for low-hours part-time employment while opening
the possibility for “foot-in-the-door” part-time employment is somewhat
conjectural. Both the FMLA and UI work tests include part-time work-
ers, although the FMLA minimum (an average of twenty-four hours per
week) seems a preferable screen for higher-hours part-time workers than
the requirement of state UI systems, whose median is about seven weeks
of employment per year.324
     At the same time, the FMLA’s requirement of stable attachment to one
employer is unduly restrictive. Recall that UI requires employment rather
than employer stability by mandating a combination of minimum hours
or earnings and some minimum degree of work-spreading through the
year. Paid family leave insurance should include workers whose work hours
are substantial and whose attachment to the workforce is stable, even if
they fail to meet the work-hours test for any single employer. Including
workers who piece together full or substantial work hours with multiple
employers seems unlikely to create signiªcant incentives for workers to
become “trapped” in such arrangements because these workers likely prefer
more stable employment and will continue to seek such opportunities. Fur-
thermore, paid leave may facilitate the move toward more stable workforce
attachment in these workers if it helps them avoid downward mobility
and family restructuring in the event of work interruption. Therefore, eligi-
bility requirements for paid leave should permit the aggregation of hours
worked across employers.
     Finally, small employers should not be exempted from the require-
ment, as they are the employers most likely to lack voluntary paid leave
programs. Furthermore, the scope of caregiving needs that trigger eligibility
for beneªts should be wider than under the FMLA. For reasons discussed
earlier, care needs of an unmarried domestic partner should be included,
rather than limiting the triggering events to the care needs of spouses,
children and parents.




     324 These are the author’s calculations based on number of weeks a full-time worker earn-

ing the minimum wage would need to work to meet the median monetary eligibility re-
quirement under state UI laws. See Lester, Wealth Redistribution, supra note 56, at 347.
2005]                 A Defense of State Paid Family Leave                              79

                      E. Incentives for Men To Take Leaves

     Among workers who have young children, women are signiªcantly
more likely than men to take leaves for an FMLA-covered reason.325 Fur-
thermore, the average length of leave taken by fathers for a birth or adop-
tion is only ªve days.326 Even if paid leave policies have the potential to
increase women’s workforce attachment—and perhaps over time alter the
balance of women’s and men’s relative contributions to work and family—
we cannot assume that women’s increased involvement in paid work, on
its own, will signiªcantly alter men’s behavior.327 Successful achievement
of this goal requires correlative, active efforts to increase men’s involve-
ment in family care.328
     The importance of this point can scarcely be overstated. If we are to
break down sex-role stereotypes about the social division of labor, in-
creasing men’s involvement in caregiving is the necessary corollary to
increasing women’s workforce participation. Absent efforts to boost men’s
participation, increasing the generosity of family leave policies could
potentially reinforce rather than reduce the gendered division of family
versus market work. I discussed earlier how the economics of the gender
wage gap skews the family allocation of market versus home work so
that women are less likely to engage in market work.329 Traditional social
pressures and corporate culture also discourage men from taking time
away from work to assume the role of a family caregiver.330 In addition,


     325 Balancing the Needs, supra note 5, at tbl.4.17 (reporting that 75.8% of women

with young children and 45.1% of men with young children took a leave for a covered
reason in 2000). Note that among all workers with young children, men were about as
likely as women to take some leave to care for a newborn or newly adopted child (34.1%
men versus 35.8% women), but this ªgure alone may be misleading because another 32.4%
of women took a leave classiªed in the separate category of “maternity-disability.” Id. at
tbl.4.19.
     326 Pleck, supra note 78, at 93–94; Selmi, supra note 1, at 755–56 (discussing the same

issue).
     327 See, e.g., Silbaugh, Turning Labor into Love, supra note 198, at 8–9 (arguing that

changes in the division of household labor as women have increased workforce participa-
tion reºect a reduction in time women devote to housework rather than any increase in the
time men devote to housework).
     328 Professor Selmi puts this point more strongly: “Increasing workplace equality will

require persuading men to behave more like women, rather than trying to induce women to
behave more like men.” Selmi, supra note 1, at 708.
     329 See supra notes 84–85 and accompanying text.
     330 Bornstein, supra note 287, at 116–17 (arguing that both workplace culture and peer

pressure discourage men’s leave-taking); Mindy Fried, Taking Time: Parental Leave
Policy and Corporate Culture 91–96 (1998) (noting that information provided by
companies, as well as attitudes of managers and workers, reºect and reinforce gendered
leave-taking arrangements); Martin H. Malin, Fathers and Parental Leave, 72 Tex. L. Rev.
1047, 1077–79 (1994) [hereinafter Malin, Fathers] (citing employer hostility and peer
pressure as factors that discourage men’s leave-taking); Martin H. Malin, Fathers and Pa-
rental Leave Revisited, 19 N. Ill. U. L. Rev. 25, 39–42 (1998) (arguing that men have
good reason to fear that seeking family accommodations will harm their careers and are
therefore secretive when they take leave for family reasons); Pleck, supra note 78, at 90
80                       Harvard Journal of Law & Gender                            [Vol. 28

under the FMLA, the “key employee” exemption, under which workers
whose earnings are in the top 10% of a covered employer may be excluded
from the job security guarantee,331 tends to exclude more men than women
from eligibility given men’s higher representation among top earners.332 If
uptake of paid leave beneªts is understood by employers to be a charac-
teristic of female workers only, rational employers may assign to young
women ºatter career proªles with less job training, fewer promotions,
and lower wages.333 I explained earlier how corrosive the cost-shifting
dynamic might be to the goal of increasing women’s workforce participa-
tion. To the extent that employers might bear at least some nominal ªxed
costs of providing the unpaid job-protection component of the beneªt,
increasing men’s uptake of beneªts is a crucial step toward countering the
cost-shifting phenomenon: the less family leave is identiªable as a “women’s
beneªt,” the less rational it will be for employers to shift the cost of provi-
sion to women.
     The European experience suggests that progress in increasing men’s
leave-taking, although not trivial, is slow.334 Most analysts agree that add-
ing a wage replacement component to existing unpaid leave programs will
increase the willingness of men to take leave because in addition to re-
ducing the opportunity cost of leave-taking generally, it will reduce the
opportunity cost differential between mothers and fathers.335 Some Euro-
pean countries have had success with more aggressive policies, such as
establishing “daddy quotas,” beneªts offered to fathers on a “use-it-or-
lose-it” basis.336 These policies condition receipt of some portion of a fam-

(involved fathers may encounter hostility from acquaintances, relatives, and workmates);
Rhode, supra note 109, at 843 (reviewing surveys of attitudes about male leaves for births
and adoptions); Selmi, supra note 1, at 758–59 (commenting that the notion of men taking
leave for childcare purposes may violate existing gender roles).
     331 29 U.S.C. § 2614(b) (2000).
     332 See Amy Olsen, Comment, Family Leave Legislation: Ensuring Both Job Security

and Family Values, 35 Santa Clara L. Rev. 983, 1017 (1995).
     333 Nabanita Datta Gupta & Nina Smith, Children and Career Interruptions: The

Family Gap in Denmark 4 (Inst. for Study of Labor, Discussion Paper No. 263, 2001);
Malin, Fathers, supra note 330, at 1062–63.
     334 See, e.g., Heather Joshi, The Opportunity Costs of Childbearing: More than Moth-

ers’ Business, 11 J. Popul. Econ. 161, 176–77 (1998) (citing British surveys for proposi-
tion that attitudes about appropriate gender roles often change more slowly than practice);
Peter Moss & Fred Deven, Parental Leave in Context, in Parental Leave: Progress or
Pitfall?, supra note 171, at 1, 13–14 (describing the challenge of encouraging fathers’
participation in paid parental leave programs in various European countries); European
Opinion Research Group, Europeans’ Attitudes to Parental Leave (European
Comm’n, Special Eurobarometer No. 189/Wave 59.1, 2004).
     335 Bornstein, supra note 287, at 116 (citing Littleton, supra note 84, at 36); Littleton,

supra note 84, at 35–36; Selmi, supra note 1, at 770–71. Cf. Malin, Fathers, supra note
330, at 1081–89 (arguing for an interpretation of existing law that would allow fathers to
combine sick leave with FMLA leave to care for mothers during postpartum recovery).
     336 See, e.g., Linda Haas & Philip Hwang, Parental Leave in Sweden, in Parental

Leave: Progress or Pitfall? supra note 171, at 46, 49 (describing Sweden’s policy of
making some portion of family leave available at 80% wage replacement to fathers on a
nontransferable basis); Arnlaug Leira, Cash-For-Child Care and Daddy Leave, in Paren-
2005]                A Defense of State Paid Family Leave                              81

ily’s paid leave allocation on the requirement that the father be the pri-
mary caregiver for that period of time. These “use-it-or-lose-it” schemes
operate against the backdrop of paid family leave being a family, rather
than an individual, beneªt, i.e., each family with a child gets a certain
allocation of paid job leave time. The unpaid leave available under the
FMLA, by contrast, is an individual beneªt, and thus is essentially already
offered on a use-it-or-lose it basis.337 The effect of wage replacement on a
paid family leave program built from this model would be to increase the
sense that the father who declines to take leave is in fact “losing” some-
thing, namely, a ªnancial beneªt. However, the proportion of wage re-
placement will have an important inºuence on the power of this incentive. In
Scandinavia, wage replacement is nearly full, making the risk of “loss”
very high for fathers who decline to take leave.
      Professor Michael Selmi has advocated making the stakes still higher,
with a proposal to provide six weeks’ paid leave to both men and women
that can be claimed only on an all-or-nothing basis by each parent.338 In
other words, each parent would lose her entitlement unless she took the
full six weeks’ leave. Presumably, a crucial assumption of Professor Selmi’s
proposal is that paid leave alone is not sufªcient encouragement for men
to take leave, and that absent an all-or-nothing stipulation, men who take
leave will on average take fewer than six weeks. This is probably correct,
although Professor Selmi’s plan may sacriªce the participation of some
fathers who would take a leave only if they could limit it to a shorter pe-
riod, perhaps two weeks or one month. An open question is which alterna-
tive—more men taking relatively shorter leaves (although doubtless longer
than we see in the current unpaid scheme) or fewer men taking longer
leaves—would do more to cultivate a norm of meaningful male participa-
tion in carework over time. It might be sensible, at least initially, to de-
termine the effect of adding a pay component without an all-or-nothing
stipulation on both the number and duration of leaves by men. If deemed
inadequate, experimentation with all-or-nothing entitlements would seem
a promising next step. Other proposals include mandating that men take
leave following the birth or adoption of a child,339 linking federal contract
eligibility to family leave uptake rates by male employees,340 marshalling
antidiscrimination law to challenge workplace hostility to men’s leave tak-
ing,341 and public education for both men and women.342 Although each of

tal Leave: Progress or Pitfall?, supra note 171, at 275–76 (describing Norway’s pol-
icy of earmarking four weeks of state-provided family leave for exclusive use of fathers).
     337 Note that where spouses work for the same employer, the employer may limit their

total permissible leave to care for a child after birth, adoption, or foster care to twelve
weeks combined. 29 U.S.C. § 2612(f) (2000).
     338 See Selmi, supra note 1, at 770–71.
     339 Id. at 773–75.
     340 Id. at 775–79.
     341 Malin, Fathers, supra note 330, at 1089–94.
     342
         Bornstein, supra note 287, at 122–23; Haas & Hwang, supra note 336, at 50 (de-
82                     Harvard Journal of Law & Gender                         [Vol. 28

these ideas merits deeper exploration, I would recommend focusing at the
very least on public awareness campaigns and further study of the likely
impact on male participation in all-or-nothing entitlements.

                              F. Duration of Beneªts

     I have argued that beneªts of excessive duration may undermine the
goal of increasing women’s workforce attachment. Examples from the
European literature of excessively generous beneªts may be found in the
three-year paid programs offered in France and Germany.343 While some
speculation is required, three to four months’ wage replacement per year
for all covered work interruptions seems more sensible. The California pro-
gram, which offers six weeks’ compensation, may reºect a more realistic
picture of what is politically possible.

                                VIII. Conclusion

     In this Article, I make the case for publicly mandated paid family
leave ªnanced in a manner that spreads its costs beyond those most likely, at
least in the short run, to take leaves (e.g., women, particularly those in
their childbearing years). Although this would mean that some citizens
who have no family caregiving needs will subsidize others who do, I be-
lieve such redistribution is justiªed. Publicly provided paid family leave
has the potential to increase women’s workforce attachment. Historically
and currently, women perform a disproportionate share of unpaid house-
hold labor, which hinders their participation in the paid workforce. Greater
involvement in the paid market labor has the potential to increase women’s
self-esteem, economic power, and sense of personal control, as well as ad-
vance social equality between men and women.
     In making this claim, I challenge two lines of argument that reject
targeted accommodation of the work-family conºict: those who argue that
on fairness grounds all caregiving labor, whether by workers or by stay-
at-home caregivers, should be equally compensated, and those who argue
that we must externalize the caregiving infrastructure to avoid reinforcing
gender segregation in the workplace. I contend that we must shift the
ªnancial incentives of caregiving labor to make continuous attachment to
paid work more attractive to caregivers than unpaid homemaking. Simul-
taneously I argue that a fully externalized caregiving policy would ignore
the speciªc demand by many workers to be able to give hands-on, per-
sonal care to members of their family.

scribing Sweden’s public information campaigns directed at encouraging fathers to take
family leaves). Note that women may also resist an increased paternal role in caregiving,
perceiving it as a threat to their own arena of dominance in the family sphere. See Cahn,
supra note 107; Malin, Fathers, supra note 330, at 1068.
     343
         See sources cited supra note 171.
2005]             A Defense of State Paid Family Leave                  83

     Some limitations on the design of paid leave policies are necessary
for them to achieve the goals I espouse. Although there is still a need for
further research, both theory and existing empirical work suggest that the
effectiveness of paid leave policies in increasing women’s workforce at-
tachment may require limiting the duration of leaves and ªnancing beneªts
in a way that avoids shifting excessive costs of the program to women.
The argument for spreading the costs of leave policies is driven in part by
the practical observation that the very incentive to enter into or remain
attached to the workforce may be thwarted if the full cost of caregiving
accommodations is borne by women workers. However, the argument for
cost spreading also has an ethical element: increasing women’s work-
force participation will advance the broader social project of reducing
gender inequality in paid employment.